Professional Documents
Culture Documents
I. UNIT -1 STATE VARIABLE ANALYSIS AND DESIGN 04 State Variable Analysis & Design
State Variable Analvsis or State Space Analvsis :-
The state variable approach is a powerful tool/technique for the analysis and design
2. UNIT - 2 STATE SPACE REPRESENTATION 24
of control system.
'The state space analysis is a modem approach and also easier for analysis using
digital computers. It's gives the total internal state of the system considering all initial
3. UNIT - 3 DERIVATION OF TRANSFER FUNCTION FROM
conditions.
STATE MODEL 42
Why do we need state space analysis?
,;i!J In' conventional methods, the systems are modelled using Transfer Function
7. UNIT - 7. PHASE PLANE ANALYSIS 105
.
approach, which is the ratio of Laplace transform of output to input, neglecting all the initial
conditions.
1. Transfer function is defined under zero initial conditions. Comparison: Classical vs, Modern Control
2. transfer function is applicable to linear time invariant systems.
Classical Control (Linear) Modern Control (Linear)
3. It is restricted to single input and single output systems. ~---------------.,-----.---------------.
~ Developed in 1920- ~ Developed in 1950-
4. Does not provides information regarding the internal state of the system.
1950 1980
5. The classical methods like Root locus, Bode plot etc. are basically trial and ~ Frequency domain ~ Time domain analysis I
error procedures which fail to dive the optimal solution required analysis & Design(Transfer and dcsign(Differential
function based) equation based)
State variables analysis can be applied for any type of systems like,
r ~ Based-on SISO ~ Based on MIMO
models models
Linear system
~ Deals with input and ~ Deals with input,
Non- linear system output variables output and state variables
Time invariant system ~ Well-developed ~ Not well-developed
- Time varying system robustness concepts robustness co~ccpts
(gain/phase margins)
- Multiple-input and multiple output system.
~ No ~
the analysis can be carried with initial conditions. Controllabi lity/Observabilit Controllability/Observabi Iit !
"
Modern Control Theory lOEE55 Modern Control Theory lOEE55
State space:-
The set of all possible values which the state vector X(t) can have( or
assume) at time t forms the statc space of the system.
Uj_ (t) ---+ ---+ Yl (t)
State l'ector:- U2(t) ---+ ---+ Y2(t)
Control
It is a (nx I) column matrix whose clements are state variables of the
U3(t) ---+ ---+ Y3(t)
system,(where n is order of system) it is denoted by X(t). System
.1Lm(t) ---+
~ Yp(t)
..
State Variable Selection
).- Typically, the number of state variables (i.e. the order of the system)
! ! ! !
is equal to the number of independent energy storage elements. However, Xt(t) X2(t) X3(t) ..... Xn(t)
there are exceptions!
Control
Is there a restriction on the selection of the state variables?
U~ ~y
System
YES! All state variables should be linearly independent and they must
collectively describe the system completely.
.' x
State Space Formulation .' The different variables may be represented by the vectors(column matrix) as
shown below
In the state variable formulation of a system, in general, a system consists of Input vector Output vector Statevariable vector
rn-inputs, p-outputs and n-statc variables, The state space representation of the
U1(t) Yl(t) Xl(t)
system may be visualized as shown in figure I, I.
Let, State variables" Xl(t), Xz(t), X3(t)' Xn{t), U2(t) Y2(t) X2(t)
U3(t) Y3(t) X3(t)
Input variables ~ Uj_ (t), U2(t), U3(t), Um(t),
'.
Modern Control Theory lOEE55 Modern Control Theory 10EE55
dxl
dt
=X
dx2 x, = ani XI + an2X2 + + annx, + bnl u, + : bnmUm
dt In matrix form the above equation can be expressed as
...Functional output equation can be written as, eras. In the first era, we have classical control theory , which deals with techniques
developed befor~ 1950. classical control embodies such methods as root locus. Bode.
Y(t) = g(X, U) Output equation ....(2)' ,
Nyquist and Routh- Hurwitz. These methods have in common the use of transfer
functions- in the complex frequency(s) domain, emphasis on the use of graphical
State Model Of Linear System
techniques, the use of feedback and the use of simplifying assumptions to approximate
);> State model of a system consist of the state equation & output equation.
the time response . since computers were not available at that time , a great deal of
);> The state equation of the system is a function of state variables and inputs as defined
equation I.
, .
emphasis. was
.
placed on developing methods that were amenable t01l'manual
by
For linear tine invariant systems the first derivatives of state ~ari~bles can be
.
compbtation and graphics. A major limitation of classical control methods was the use of
single -input, single output (SISO) methods .. Multivariablc (i.e multiple - input
expressed as linear combination of state variables and inputs.
multiple - output or MIMO) systems were analyzed and designed one loop at a time
XI = all X] + al2 X2 + + aln Xn + bll Ul +.., + bIro Urn Also the usc of 1iransfer functions and the frequency domain limited one to linear time -
invariant syst;ms.
X2 = a21X] + a22X2 + + a2nXn + b2l Ul + + b2rnUrn
" In't!'le seoond era, we have modern control ( which is not so modern any longer
representation results in a system description interms of 'n' first -order differential f(t) and the natural or unforced response ie the solution of homogeneous equation
equations, as opposed to the usual nth -order equation. s A convenient tool for X+aX+bX=O
this new system representations is matrix notation. If XI (t), X2(t) -------- Xn(t) are state variables of the system chosen then the
System state and state yariable initial conditions of the state variables plus the u(t)'s for t > 0 should be sufficient to
It is important to stress at out set that the concept of system state is first of all, a decide the future behaviour i.e y(t)'s for t > O. Note that the state variables need not be
physical concept. However it is often convenient to interrns of mathematical model. physically measurable or observable quantities. Practically however it is convenient to
Here this mathematical model is assumed to consist of ordinary differential equations -----Chose easily measurable quantities. The number of state variable is then equal to the
which have a unique solution tor all inputs and initial conditions. It is in terms of this order of the differential equation which is normally equal to the number of energy
mathematical model that the 'system state' or simply 'state' is defined. storage elements in the system.
X I (to) ,X,(t,,) -------- X,,(to) which along with the input to the system for t to is State eguatjon~ for linear systems in matrix form
sufficient to determine the behaviour of the system tor all t to. The state of a linear time -invariant nth order system is represented by the following set
of 'n' number of first order differential equations with constant coefficients in terms of n
state variable XI,X2-------- X,
Output eQuatjon
b"T
The state variables XI(t) -------X net) represents the dynamic state of a system. The.
o
X A X + B u Cp1 Cp2 ----------- Cpn
dn-Iy
dy,
yeO), -t6t
Cpn OP1 OP2 --------- QPm ------------ - (0), are initial conditions
.dt' dt
The state equation of a system determines its dynamic state and the output equation gives We first define the state variables
its output at any time t > 0 provided the state at t = 0 and the control forces for
t 0 are known. These two equations together form the state model of the-system, XI, X2: ---------- Xn which can be done in many possible ways, A convenient way
The state model of a linear system is therefore given by is define'
X=AX+BU (I)
Y = CX+ DU
..
State Model of SISO linear and time invariant svstem,
If we let m = I and p= I in the state model of a multiple input multiple output linear time
x, =y",1
invariant system we obtain the following state model for SISO linear system. With the above definition of state variables equation (4) is reduced to a set of'n' first
order differential equations given below;
x AX +bu
0
----+(2)
XI=Y =X2
Y=CX+du X2 ='1 = X3
. ,
I 'I
Xn_1 = l-I = X,
the above equations result in the following state equations Deriyation of transfer functjon from a given state model
0 1 0---------- .- 0 Xl 0 Having obtained the state model we next consider the problem of determining transfer
0 0 1 ------------ 0 )\2 0 function from a given state model ofS[SO I MIMO systems.
0 0 1 ----------- 0 .'
I) 'SISP SYSTEM
0 0 0 ------------- 1
-~ -al -a2 ------------ -an x,
+
0
bo
u
u(s)
;_;__----.l. I' G( s)
yes)
It is to be noted that the matrix A has a special form. It has all 1's the upper off G(s) is called transfer function defined as
- diagonal, its last row is comprised of the negative of the coefficients of the original yes)
differential equation and all other elements are zero. This form of matrix A is known as G(s) = yes) = u(s) G(s) or y (s) = G(s) u(s) -------- (I)
u(s)
the 'BUSH FORM' .The set of state variables which yield 'BUSH FORM' for the
matrix A are called 'Phase variables' State Model is given by
From (4)
(sI -A) Xes) = Bu(s)
Substituting ( 6) in (5)
)',(s) = G , Irs) lI,(s) + G ds) U2(S)'1 -------------------+_ G Im(S)Um(s) X3+ 9X3+26 X2+ 24 X1=24R
)'I(S) Transfer function
X3~ - 24XC26 X2-9X3 + 24R
lIes) =
C(S)
C= [1 X1(S)
C(S)
R(S)
3. Cascading form
---- _ ~--~~------
X1(S)_ 1
83 + 982 + 26s +24
The denominator ofTF is to be in factor fOIl11
C(S) 24
Taking in LT R(S)
24
(s +2) (S+3) (S+4)
Xl
Xl=X2
Xl =X2= X3 X
X~
[ X3
2 [0 1 0J [X~rOl
0
-24 -26-9
1 X
X3
+ 0 R(t)
1
X'l = X3
'. Xl 1
----
s+4
= _ _;;_---
1 + 4 s-1
s-1
X2
.'
Modern Control Theory
lOEE55
Modern Control Theory lOEE55
UNIT-2
STATE-SPACE REPRESENTATION
1 Introduction
Xl=X2-4Xl The classical control theory and methods (such as root locus) that we have been using in
class to date are based on a simple input-outputdescriptionof the plant, usuallyexpressedas a
X2=X3- 3 x r transfer function. These methods do not use any knowledgeof the interior structure of the plant,
X3= - 2X3+24V and limit us to single-input single-output(SISO) systems,and as we have seenallowsonly limited
control of the closed-loopbehavior when feedbackcontrol is used. Modem control theory solves
manyof the limitationsby using a much"richer"descriptionof the plant dynamics. The so-caned
C = [1 o state-space descriptionprovide the dynamicsas a set of coupled first-order differential equations
in a set of internal variables known as state variables, together with a set of algebraicequations
that combinethe state variablesintophysicaloutput variables.
- ----rhe concept of the, state of a dynamic system refers to a minimum set of variables, knownas
state variables, that fully describe the systemand its responseto any given set of inputs[1-3]. In
particulara state-determined system modelhas the characteristic that:
A mathematicaldescriptionof the systemin terms of a minimumset of variablesxi(f), i
= 1... n, !ogetherwith knowledgeof those variablesat an initial time 10 and the
system inputs' for time t 2:: to, are sufficientto predict the future system state and
outputs for alltime t > to.
[Type text]
Page 24
Dept. of EEE.SJBIT
Page23
"
Modern -Control Theory 10EE55
Input vector U Output vector y where Xi = dx./dt and eachof the functions ,Ii (x, u, I), (i = 1. ... , n) may b",.a general
nonlinear, time varying function of the state variables, the system inputs, and time. I
Uf:,"(I) ~ System
...~ descibed by state variables
-L..rl.> l)
It is common to express the state equations in a vector form, in which the set of 11 state
variables is written as a state vector x(t) = [XI (I), X2 (t). ' .. ,x,,(t)(, and the set of I' inputs
is written as an in~)Utvector U(I) = [1/, (I), U2 (f), ... , u,. (t)f. Each state variable is il time
u~(I) ~,--_{X_1 ,X_2_'
,_
.. ~_}--'t Ym(t) " varying component of the column vector X(/).
This form of the state equations explicitly represents the basic elements contained
Figure 1: System inputs and outputs. in the definition of ,a state determined system. Given a set of initial conditions (the values
of the Xi at some time (0) and the inputs for I ?: 10, the state equations explicitly specify
the derivatives of all state variables. The value of each state variable at some time f;.1 later
system models. For suchsystems the number of state variables, n, is eflual to the number of may then be found by direct integration. The system state at any instant may be
independent energy storage elements in the system. The values of the state variables at any interpreted as a point in an l1-dimensionalstate space, and the' dynamic state response x(t)
time t specify the energy of each energy storage element within the system, and therefore can be interpreted as a path or trajectory traced out in the state space.
the total system energy, and the time derivatives of the state variables determine the rate In vector notation the set of 11 equations in Eqs. (1) may be written:
of change of the system energy. Furthermore, the values of the system state variables at
any time t provide sufficient information to determine the values of all other variables in the . ,
"
x = f (x. u, I). (2)
system at that time. . ,
where f (lI.. u. I) is a vector function with 11 componentsr, (x. u, I). ,
There is no unique set of state variables that describe any given system; many In this note we restrict attention primarily to a description of systems that arc linear and
different sets of variables may be selected to yield a complete system description. lime-invariant (LTI), that is systems described by linear differential equations with constant
However, for a given system the order n is unique, and is independent of the particular set coefficients. For an,LTI system of order II, and with I' inputs, Eqs. (1) become a set of 11
of state variables chosen. State variable descriptions of systems may: be formulated in coupled first-order linear differential equations with constant coefficients:
terms of physical and measurable variables, or in terms of variables that are {lot directly , XI + + + ([,,,X,, + blltl, + + o.,;
measurable. It is possible to mathematically transform one set of state variables to , X2
,
,;- t(l11X, G12X2
6 a21XI + G22X2 + ." + G2nX" + b21UI + ." + h21'U,.
another; the important point is that any set of state variables must provide a complete (3)
description of the system. In this note we concentrate on a particular set of state
variables that are based on energy storage variables in physical systems: Xn O"IXI + an2X2 + " , + QnnXIl + bnlUI + ". + hl,:r
where the coefficients aij a~d bij are constants that describe the system. This set of 11
1.2 The State Equations equations defines the derivative~ of the state variables to be a weighted sum of the state
variables and the system inputs.
A standard form for the state equations is used throughout system dynamics. In the standard Equations (8) may be written compactly in a matrix form:
form the mathematical description of the system is expressed as a set of n' couple'd first-order
ordinary differential equations, known as the stale equations, in which the time derivative As given above in page 10
of each state variable is expressed in terms of the state variables XI (r),... , x; (I) and the
system inputs UI (I), ... , Ur (t). In the general case the form of the 11 state equations is: x = Ax +Bu (5)
In this note we use bold-faced type to denote vector quantities. Upper case letters arc
XI = fi (x,u, t)
used to denote general matrices while lower case letters denote column vectors. See
X2 = fi(x, u, t) Appendix A for an introduction to matrix notation and operations.
(1)
x; =./;' (x, u, t
[Type text] Page 25 [Type text] Page 26
"
"
Modern Control Theory lOEE55 Modern Control Theory lOEE55
where the state vector x is a column vector of length II, the input vector u is a .column vector
of length r, A is an 11 x 11
. '. .
square matrix of the constant coefficients ai;, a'nd B is an n x r
1.4 State Equation Based Modeling Procedure
The ~omplet~system model for a linear time-invariant system consists of (i) a set of II
matrix of the coefficients b" that weight the inputs. state equations, defined in terms of the matrices A and B, and (ii) a set of output
1.3
.. .
Output Equarioris
equations th'at relate any output variables of interest to the state variables and inputs,
and expressed in terms of the C and D matrices. The task of modeling the system is to
derive the elements of the matrices, and to write the system model in the form:
A system output is defined to be any system variable of interest. A d~crtpti(;m of a Ax +Bu (11)
y Cx + Du. (12)
physical system in terms of a set of state variables does not necessarily include all of the
variables of direct engineering interest. An important property of the linear s\ate 'equation The matrices A and B are properties of the system and are determined by the system
structure and elements. The output equation matrices C and D are determined by the
description is that all system variables may be represented by a linear combination of the particular choice of output variables.
The overall modeling procedure developed in this chapter is based on the following steps:
state variables Xi and the system inputs u.. An arbitrary output variable in a system of
1. Determination of the system order n and selection of a set of state variables from
order 11 with r inputs may be written: the linear graphsystem representation.
(6) 2. GeAeratioh of a set of state equations and the system A and B matrices using a
well ~fined! methodology. This step is also based on the linear graph system
where the c, and d, are constants. If a total of III system variables are'defined as outputs, description.
the 111 suchequations may be written as:
y] 3. Determination of a suitable set of output equations and derivation of the appropriate
ClIX] + ('12.\2 + + ell/X" + d]]u] + + d1ru,. C and D matrices.
)'2 ['21Xl + ('22.\"2 + + e21/XII + d2]1I] + + (hrllr
(7) 2 Block Diagram Representation of Linear Systems
Described by State Equations
or in matrix form:
The matrix-based state equations express the derivatives of the state-variables explicitly in
As given above in page 10 terms of the states themselves and the inputs. In this form, the state vector is expressed as
the direct result of a vector integration. The block diagram representation is shown in
Fig. 2.. This general block diagram shows the matrix operations from input to output in
.-
The output equations, Eqs. (8), are commonly written in the compact form:
terms of the A, B, C, D matrices, but does not show the path of individual variables .
y = Cx +Du (9) In state-determined systems, the state variables may always be taken as the outputs of
integrator blocks. A system of order II has II integrators in its block diagram. The
where y is a column vector of the output variables y;(t), C is an III x II matrix of the.constant derivatives of the state variables are the inputs to the integrator blocks, and each state
coefficients C;j that weight the state variables, and D is an III x r matrix of the constant equation expresses a derivative as a sum of weighted state variables and inputs. A detailed
coefficients ,i;, that weight the system inputs. For many physical systems the matrix D block diagram representing a system of order n may be constructed directly from the state
is the null matrix; and the output equation reduces to a simple weighted combination of and output equations as follows:
the state variables:
y =Cx. (10) Step I: Draw n integrator (S-] ) blocks, and assign a state variable to the output
of each block.
[Type text]
Page 27 [Type text]
Page 28
Model II COIILI 01 Theory lOEES5 Modern Control Theory lOEE55
'--
JL
--,I,.~ ~
ro -v
"j
B W':+-+":"'
"
I~r"'
'~_"_j:--....,)XI SA,_I II.,:.,X,:_" '>_"
r, ,. __1 .)=;> YI'J
~.;..(+, u(l)
'. +
+v-->y(t)
Figure 2: Vector block diagram for a linear system described by state-space 'system dynamics ..
+
Step 2: At the input to each block (which represents the derivative of its state variable)
draw a summing element.
Step 3: Use the state equations to connect the state variables and inputs'to the
summing elements through scaling operator blocks. .'
Step 4: Expand the output equations and sum the state variables and inputs through
a set of scaling operators to form the components of the output. ' Figure 3: Block diagram for a state-equation based second-order system .
Example 1 Example 2
Draw a block diagram for the general second-order, single-input single-output Find the transfer function and a single first-order differential equation relating
system: the outputy(t) to the input l/(t) for a system described by the first-order linear
XI state and output equations:
XI all al2
+
hi
X2 u(t) dx
X2 a21 a22 b: ax(t) + bu(t) (i)
XI
dt
y(t) CI C2 + du(t). (i) y(t) cx(t) + du(t) (ii)
X2
Solution: The block diagram shown in Fig. 3 was drawn using the four steps Solution: The Laplacetransform of the state equation is
described above. '
~X(s) = aX (s) + bU(s). (iii)
3.Transformation From State-Space Equations to Clas- which may be rewritten with the state variable X(s) on the left-hand side:
sical Form (s - ol X)) - bl1l-----c- _ (iv)
The transfer function and the classical input-output differential equation for any system Then dividing by (s - a), solve for the state variable:
variable may be found directly from a state space representation through the Laplace
transform. The following example illustrates the general method for a first order system. b
X(s) =s _ a U(s). (v)
and substitute into the Laplace transform of the output equation Y (s) = cX(s) + The state equations, written in the form of Eq. (16), are a set of n simultaneous
dUes): . opera- tional expressions. The common methods of solving linear algebraic equations, for
example Gaussian: elimination, Cramer's rule, the matrix inverse, elimination and
be
Yes) --+d U(s) substitution, may be directly applied to linear operational equations such as Eq. (16). For
.'i-a
low-order single-in~ut single-output systems the transformation to a classical formu- lation
-=-._ ds +U(s) (vi) may be performed 'in the following steps:
(be - ad) .. :
(s - a) 1. Take t~e Laplace transform of the state equations.
The transfer function is: "
Yes) d\+ (be- 2. Reorganize each state equation so that all terms in the state variables are on
fl(,) =-- =
. U(s) ad) (8 - or (vii) the left-hand .side,
The differential equation is found directly: 3. Treat the state equations as a set of simultaneous algebraic equations and solve
for those state variables required to generate the output variable.
(.\'- a) Y (s) = (ds + (be - ad)) U(s). (viii)
4. Substitute for the state variables in the output equation.
and ret\'rif;ng as a differential equation:
dv du 5. Write the output equation in operational form and identify the transfer function.
dt - ay = d dt + (be - ad) u(l). (ix)
6. Use the transfer function to write a single differential equation between the
Classical represenfations of higher-order systems may be derived in an analogous set of output variable and the system input.
steps by using the Laplace transform and matrix algebra. A set of linear state and output This method is illustrated in the following two examples.
equations written in standard form
Ax +Bu (13)
Example 3
y ex +Du (14) Use t~e Laplace transform method to derive a single differential equation for the
capacitor voltage Vc in the series R-L-C electric circuit shown in Fig. 4
may be rewritten in the Laplace domain. The system equations are then
Solution: The linear graph method of state equation generation selects the
sX(s) = AX(s) + BU(s)
Yes) = CX(s) + DU(s)
~ (16) "s(t)
.~/~ 1(1 R C
"
where the term "I creates an II x II matrix with s on the leading diagonal and'zeros elsewhere.
(This step is necessary because matrix addition and subtraction is only defined for matrices
of the same dimcnsion.) 111C matrix [sl - A] appears frequently throughout linear system
~_7
theory; it is a square II x II matrix withclemcnts directly related to the A matrix: (8) Physical system (b) Linear graph (c) Normaltree
c
(.\' - (/11') -a12
-all (05- (/22) -a2n
[.1'1- A] ~ (17) Figure 4: A series RLC circuit.
-(1;,1 (s - a",,)
:
.
Model II Corm 01 Theory 10EE55 Modern Control Theory 10EE55
capacitor voltage vc(t) and the inductor current iL(t) as state variables, and Cramer's Rule, for the solution of a set of linear algebraic equations, is J useful method to
generates the following pair of state equations: . apply to the solution of these equations. In solving for the variable Xi in a set of 11 linear
algebraic equations, such as Ax = b the rule states:
o l/C (i)
-1/L -R/L det A(i)
Xi = det[A] (IS)
The required output equation is:
-
where A (,) is another n x n matrix formed by replacing the ith column of A with the vector
y(t) = 1 0 (ii) b.
If
Step I: In Laplace transform form the state equations are: [.1'1- A] X(s) = BU(s) (19)
sVc(s) = OVc(s) + IIC h(s) + OVs(s) then the relationship between the ith state variable and the input is
sh(s) = -lILVc(s) - RlLh(s) + 1ILI/,(s) (iii) det lsI - A](i)
Xi(S) =--------U(s) (20)
det lvl - A]
Step 2: Reorganize the state equations:
sVc(s) - IIC h(s) OVs(s) (iv) where (sl - A) (i) is defined to be the matrix formed by replacing the ill,' column of (sJ - A)
with the column vector B. The differential equation is
1ILVc(s) + ls+ RILlh(s) lILV,(s) (v)
det [vl - A],, = det (.I'I - At) Ilk (/). (21)
Step 3: In this case we have two simultaneous operational equations in the
state variables vc and it.. The output equation requires only Vc. If Eq. Example 4
(iv) is multiplied by [.I' + RILl, and Eq. (v) is multiplied by IIC, and the
equations added, h(s) is eliminated: Use Cramer's Rule to solve for Vl.(t) in the electrical system of Example 3.
Solution: From Example 3 the state equations are:
[.I' (s + RlL) + IILC] Vc(s) = IILCVs(s) (vi)
vc o I1C
+
o (i)
Step 4: The output equation is y = Vc. Operate on both sides of Eq. (vi) by -lIL -RIL IlL
[S2 + (RlL)s + IILCrl and write in quotient form:
and the output equation is:
_ IILC v: (vii)
Vc(s) - S2 +(RIL)s + 11LC s(s) VL = -Vc - RiL + V,(t). (ii)
Step 5: The transfer function H(s) = Vc(s)IV,(s) is: In the Laplace domain the state equations are:
H(s) -liTe
S2 +(RlL)s+I1LC s -IIC V,.(s) o
(viii)
Vi"(S), _(iii)
IlL s +RIL h(s) IlL
Step 6: The differential equation relating vc to V, is:
Since
s+R/L l/C
o
C adj (sI - A)B -1 -R
-1IL s
IlL
..
R 1
-r -Le' (vi) .
which is the same result found by using Cramer's rule in Example 4. Figure 5: Block diagram of a system represented by a classical differential equation.
and rearranged to generate a feedback structure that can be used as the basis for a block
(32)
diagram:
The block diagram provides a convenient method for deriving a set of state e({llationsfor '1 1,,-1 1 aI I au 1
2(s) =-U(s) - -- + ... + -;=T + --_-" 2(s). (33)
a system that is specified in terms of a single input/output differential equatio,n. A set of an an s an s an s
n state variables can be identified as the outputs of integrators in the diagram, and state The dummy'variable 2(s) is specified in terms of the system input lI(t) and a weighted slim
equations can be written from the conditions at the inputs to the integrator blocks (the of successive integrations of itself. Figure 5 shows the overallstructure of this direct-form
derivative of the state variables). There are many methods for doing thi,s; we present block diagram. A string of 11 cascaded integrator (lis) blocks, with 2(s) defined at the input
here one convenient state equation formulation that is widely used 'in control system to the first block, is used to generate the feedback terms, 2(s)ls-l, i = 1,... 11, in Eq, (33).
theory. : Equation (3l) serves to combine the outputs from the integrators into the outputy (I).
Let the differential equation representing the system be of order n, and withou,t loss of A set'of.-state equations may be found from the block diagram by assigni})gthe state
generality assume that the order of the polynomial operators on both sides Ps the same: variables ,x,(t) to the outputs of the Il integrators. Because of the direct cascade connection
a"s" + a"~ls"-l+ "'+ao Y(s)= b"s"+b"~IS"~l+ "'+bn U(s). (28)
of the integrators, the state equations take a very simple form. By inspection:
We may multiply bothsides of the equation by s?" to ensure that all differenjial operators
have been eliminated:
a" + a"~ls~1 + ...+ als~("~I) +lOS~" Y(s) =
"
b"+b"~IS~1 +"'+bls-("~I)+"'+bos~" U(.t). (29) U() (II llfl~ I 1
--X1 - -X2 .. - ---x" + -~-II(I). (34)
from which the output may be specified in terms of a transfer function. If we define a dummy nit all (in (In
o where H(s) is defined to be the matrix transferfunction relating the output vector Y(s) to
o + 1 Ollu(t), (ii) the input vector U(s):
I
-13 -19 -7
[Type text] Page 39 [Type text] Page 40
Modern Control Theory lOEE55
Modern Control Theory 10E~55
"
H(s) = (C adj (sl - A) B + det lsI - A] D) UNIT-3 DERIVATION OF TRANSFER FUNCTION FROM STATE MODEL
(41)
det[sI-A]
UNIT-3
Derivation of transfer function from state model, digitalization, Eigeri values, Eigen vectors,
generalized Eigen vectors.
6 Hours
The state variables that produce a state model are not, in general. unique. HO\~eVeLthere ex ist several
common methods of producing state models from transfer functions, Most control theory texts contain
developments of a standard form called the control canonical form, sec, e.g., [I]. A:;ther is the phuse
variable canonical form.
When the order of a transfer function's denominator is higher than the order of its numerator, the
transfer function is called strictly proper. Consider the general, strictly proper third-order transfer
function
"
.r.<!l = /),.\"'+ h,s + h" (1)
U(s) .1'] + a,.\"' +<I,s+a"
"
Dividing each term ~y the highest order of s yields
Y(s)
(2)
U(s)
which is a function containing numerous lis terms, or integrators. There arc various signal-now graph
configurations that will produce this function. One possibility is the control canonical form shown in
Figure I,The state model for the signal flow configuration in Figure I is
.1'~["" ", b,l~:l+[Olll which is identical to equation (2) proving that the control canonical form is valid.
Lx,
As an example consider the transfer function
Y(s)
C(s) 2s'+8s+6
R(s) s'+8s'+26s+6
I. I I I I I
Yes) +(1, -'} (s)+<I, ,Y(s)+ao,Y(s) = b, -U(s) +b, ,U(s)+ho,U(s) (4)
s s, S s s s
and then substituting the expression for Yes) from the state model output equation in (3) yields
, " .. Examination! of Figure 1 shows one potential benefit of manipulating systems to conform to control
h"X,(s) + ",X,(s) + h,X,(s) =1>, ~U(s) + b, -',U(S) +h" ~U(S) canonical form. If xtCs) happened to carry units of inches (position), then X,es) might be inches/second
s s s
(veloclrv), and X,es) might be inches/second/second (acceleration),
I
-lI, -(boX,(s)
s
+ b,X,(s) + b,X,(s)
..
Phase Variable Canonical Form
- (/, -',(""X,(s) + h,X,(s) + h,X,(s)) Devl!lopme~t of the phase variable canonicalform as presented in [2J begins with the general fourth-
s order, strictly proper transfer function
1
- (/" ,(b"X,(s) + h,X,(s) + h,X3(S).
s Yes) b,S3 + b,s' +b.s + bo b,s -I + b,s -2 + bls -3 + bos"
The denominator in (6) is fourth order and leadsone to concludethat a block diagramconsistingof
four l/s terms may be useful.Constructionof the phasevariablecanonicaiform is Initiated by setting
the output to SAMPLEDDATAAND DIGITAL CONTROL SYSTEMS
(8)
which maybe representedusingfour integrators asshownIn Figure2. Figure 2.1 (a) shows the blockdiagram of a sampled-data control system. The continuous error'signal ell) is sampled at an
interval of time T by means of a sampler. The plant output is also a continuoussignal and it is due to the het that a hold
circuit preceeds the plant. -
It is important to note that a particular transfer function may be represented by block diagrams of
many different canonical forms, yielding many different valid state models. .
(tl)
The advances made in microprocessors, microcomputers, and digital signal processors have Figure 2.1:. (a) Blockdiagram of a sampleddata control system. (b) Blockdiagram of a digital control
accelerated the growth of digital control systems theory. TIle discrete-time systems are dynamic
system
systems in which the system-variablesare defined only at discrete instants of time.
The terms sampled-data control systems, discrete-time control systems and digital control Figure 2.1 ~) shows the blockdiagram of a digital control system. A digital control system uses digital Signals and a digital
systems have all been used interchangeably in control system literature. Strictly speaking, sampled- compuler to control a process. The continuoustime errorsignals are converted from analog form to digital form by means of an
data are pulse-amplitude modulated signals and are obtained by some means of sampling an analog AID converter. In practice, the AID converter itself contains the sampler. After processing the inputs, Ihe digital computer
signal. Digital signals are generated by means of digital transducers or digital computers, often in pro~des an output in digital form. This output is then converted to analog form by means of a LlIA convertel In practice, the
digitally coded form. The discrete-time systems, in a broad sense, describe all systems having some DlA converter rtself contains the holddevice. ...
form of digital or sampled signals.
Discrete-time systems differ from continuous-time systems in that the signals for a discrete-time The use of sampled data in control systems enables time.sharing among different input signals using the same control
systems are in sampled-data fonn. In contrast to the continuous-time system, the operation of equipment. Different input signals can be sampled periodically by staggering sampling time and thus a nurnber of inputs can
discrete-time systems are described by a set of difference-equations. The analysis and design of be handled overthe same control equipment.
discrete-time systems may be effectively carried out by use of the a-transform, which was evolved
from the Laplace transform as a special form. Sampled data technique is most appropriate for control systems reqUiringlongdistance data transmission. It is well known
that pulses may be transmitted with little loss of accuracy. Data in analog form mal' suffer considerable distortion in the
~-d Iii
Figure 2,2 (c) shows the Signal.modulated pulse train having pulse-width M and is a case of practical.sampling, However,
Sampler
in the case of ideal.sampling, the pulse width t::.tapproaches to zero and therefore the output ((I) of an ideal sampler is the
signal.modulated impulse train as shOWl in Figure 2.2 (~,
A. sampler is the basic element of. discretetime conlrol system, which converts a continuoustime signal into a train of
, ,
pulses occuring at the sampling instants 0, T,2T, '''' where T is the sampling~d, Note that, in between the sampling
instanls, the sampler does not transmil any information, Figure 2.3 shows two samplers operating in synchronism is., their sampling time and instants of operation are the same.
The continuoustime signal f(l) is thus being sampled twice, As the two samplers operate in synchronism, the output of the
second sampler will ob~ously be r(Q.
Figure 2.2 (a) shows a switch being used as a S31l111lerA. continuouslime signal f(I), as shown!n Figure 2.2 (b), is the input
to the sampler, The switch is closed for a very short duralion of time t::.1, and then remains open for some dural ion of time,
f(t)~[r(~J'.r(~
T T
SampieH Sampler2
Figure 2.3: Signal f(l) sampled twice
Therefore
~!
.
[I (t)] =1..'(t)-.
"
"
Thus, starring a starred function results in the starred function ITself
tT"
-)f~
Sampling Process
-H-.t
--r~.
1'1 The sampler converts a continuoustime signal W into a sequence of pulses 'Whereinthe magnITudeof the pulse gives the
Figure 2.2: Uniform penodic s.mpllng: (a)Sarnpler (b)Continuous signal (c)Practical sampling (0ldeal strength of the input signal at the instant of sampling, The pulse considered is of very short time duration and thus may be
sampling approximated as an impulse,
Considering the sampler output to be a train of weighted impulses, we can relate the continuoustime signal f(Q to the
sampler outputt'(t) as
Dept. of EEE, 5J8fT Page 47 Dept. of EEE, 518fT
Page 48
Modern Control Theory 10EE55 Moderri Control Theory 10EE55
!XI
eo
(2.4)(1) = L f(kT)o(t-kT).
k=O
, I,!
. ~
~~
-':;;~.!vJt.1t'.l':)r -1rl1d."
'-6'.,ff~rt..tf"t. T 1<]1",1) I
'.
2,2,2,1 Laplace Transform of a Sampled (Starred) Function
1,2, ...... , is the firsftimederwative of the sampled unt-step function u(l k~ occuring at t = kT; k = 0,1.2, ...
'.
Dept of EEE, SIB!T Page 49 Dept. of EEE, SIB!T Page SO
Modern Control Theory lOEE55 ModerrcContro! Theory lOEE55
"X
F '(5) is known as the Laplace transform of a sampled (starred) function or simply a starred transform.
..
If f(l) is a unit-step function, Ih~/) f(k D = 1 and the Laplace transform of the starred unit-step function r(l) is given by
F (1)
(2.6)F" (1)
t I.e-skT
'Xi
k=:()
Figure 2.5: (a)Sampler (b)Sampler input Fourier spectra (c)Sampler output Fourier (a)Sampler output Fourier spectra: w s
_1_
F (I) .I_e-fl .
~ is immediately obselVed from Figure 2.5 (c) and Figure 2.5 (d) that so long as w s ~ 2 w m' the original spectrum is preserved
in the sampled signal and can be extracted from it by low-pass filtering (shown in the dolled line in Figure 2.5 (c)). This is
Signal Reconstruction: Holding Devices the well-known Shanon's sampling theorem according to VYflichthe information contained in a signal is fully preserved in the
sampled version so long as the sampling frequency (w ~ is at least twice the maximum frequency (w rJ contained in the
The use of a sampler in control systems generates high-frequency components in the sampled oupu!. The sampler Input signal. That is,
Feurler-specna is shown in Figure 2.5(b) wherein wm is the highest frequency contained in the input signal. The sampler
output contains the Originalinput frequency component wm and complementary high-frequency components in addition, as
shovln in Figure 2.5 (el and Fi[IUre2.5 (d) that correspond to W s > 2w m and OJ s < 2w m respectively, where w s is the
sampling fre(IUency. The original signal is reconstructed from the sampled signal by means of various types of hold-circuits (extrapolators). The
simpfest hold-circuit is the zero-order-hold (ZOH) in which the reconstructed signal acquires the same value as the last
received sample for the entire sampling period. The zero-order-hold (ZOH) circuit has the characteristics of a low-pass filter and
f(J,J)~F'(jW)
therefore, if a ZOH circuit is placed after the sampler, the high-frequency (complementary) components present in the
T (al recMSlftJcted signal are easily filtered out and at the output of ZOH the original signal appears. The ZOH de~ce is also known
as a "box-car" generator. The schematic diagram of sampler and ZOH is ShOVv11 in Figure 2.6 (a), while signal reconstruction
is illustrated in Figures 2.6 (b) and 2.6 (c).
f'(t)
. :. ..
Therefore, the Laplace transform of the output of ZOH is given as
1 1 -sT
H(s) S- s' e
, (I-e....n
If(s) = ---s
'.,
Now, the transfer function of ZOH is given by
.'
Put s = jlJJ in the pteceeding relation and we have
(l-e -j"T)
Figure 2,7: Unit impulse input to ZOH jw
h(t) = u(t) - u(t - T).
(2.8B)
()
O' sin( :.':) > 0
I -J..;T12
\e -e -l-,rr") . T!~ { 7r, sin(~} < 0
:::
p
. ' ,e jw ~
To view th: previou~derivations,the amplitude and phase plots are shown in Figure 2,8.
Ia,.,{jwll
S
. (' IT)
"Ill ,-.
u-
t
iA!j"m
--:;;-
"
Te-pTiZ '( wT)
(1) . Sill T
0)
~ w, 3;0.
l
LGI>cIi<
1': 'bT)
sU'T' -jwT12 3""
(1) ,e , , .:
If the sampling frequency is (Ij m- then the sampling time T is given by -180' " ; "" " "" ..
T= ,b1..
f.;.,,!s
'.
. (b)
Figure 2.8: Frequency response of the IOH: (a)Amplitudeplot (b)Phase plot
Gho(j(u) (6
= Tsin(~) e-jC~;).
sin("W)
(2.8A)IGho(j(u)I == T (ii
Dept. of EEE. Sl8fT
Page 55 Dept. of EEE, 518lT
Page 56
Modern Control Theory lOEE55
Modern Control Theory lOEE55
After obtaining the various mathematical models such as physical, phase and canonical at 1=0 is driven to a state X(I) at time 't ' by matrix exponential function e " . This matrix
forms of state models, the next step to analysis is to obtain the solution of state equation. From the exponential function, which transfers the initial state of the system in '[' seconds, is called
solution of the state equation, the transient response can then be obtained for specific input. This STATE TRANSITION MATRIX (STM) and is denoted by
completes the analysis of control system in state-space.
Properties of State Transition Matrix (I):
The solution of state equation consists of two parts; homogenous solution and forced solution. The
1). (0)=1 proof follows from the definition,
model with zero input is referred as homogenous system and with non-zero input is referred as forced
e
A,
= 1 t At t --
iI't' A'I'
t -- t
system. The solution of state equation with zero input is called as zero input response (ZIR). The solution 2! 3! .
of state model with zero state (zero initial condition) is referred as zero state response (ZSR). Hence = ~(/)
total response is sum of ZIR and ZSR. Substituting t=O results into (0)=1.
I'
1--+-+
2 3
I'
...
,
1-1- + -t
I'
2
..
,
~(/) =
Example:
. -I [SI - A]-'
.
, (3 2 J
_I + 1.- -_ e te-I_,_ te-I Obtain the STM by Laplace Transform (Inverse Laplace Transform) method for given
2 -, T
system matrices
+ I I
a). A = , ' o 1 b). A=
o
Dept. of EEE, SjBIT
Page 59 Dept. ofEEE, SjBIT
Page 60
'.
Modern Control Theory lOEE55
Modern Control Theory lOEE55
-I
1 o o S -I
c). A = b).A=_1_2 [SI- A] =
S-I - 2 ' (S ~ 2)
-I
2 3
S+2
a). A = 0 lSI - A] = - 1 -,
OS-I S (S + I)' (S + I)
~{t) = [SI - Ar' =
S+2 -[ S
S-I
(S+ I)' (S+I)'
S-I -I 0 S -I
~(s) = lSI - Ar' = '" (I + t)e" te"
o S -I [S_1]2 "'( )
't' t = .l't'(t) =
-Ie -, (I - t)e -,
e' te'
~(t) = .1~(S) = 0
e o I
lSI _ A] = S -I
c). A = _ 2
-3 2 S+3
S+3
., lSI - Ar' =
S -I -2 S
2 S+3 (S + 2)(S + I)
(S + 2)(S + I)
- 2
I
S ..
S+3 I
- 2 S
~(t) = .'~(I) = ., (S + 2)(S + I)
2e-' - " e' - e-.:!I
- 2e-' + 2e-21 - e' + 2e--:'1
.
= ~(t)
where rxo ,rx, .. ,rx, constant coefficients which can be evaluated with eigen values of
. ~...
matrix A as described below.
.
~: For a given matrix, form its characteristic equation IAi- AI = 0 and find
eigcn values as A" A, ,..... 1., .
'.
Modert._1Control Theory lOEE55
Modern Control Theory lQEE55
Step2 (Case 1):lfall eigen values are distinct, thcn form 'n' simultaneous
equations as, I(A)=A'O = ()(ol t()(,A
I 0 0
= ()(.o O +,
- I -2
()(o 0 0 ()(,
+
e'''' "I(A,,) = ()(0 t ()(, A" t IX2 A,,' t ..... 0 ()(o -()(, - 2()(,
()(n -()(,
11;:
= l = e' (1- I)
Hence STM is given by
Since it is case of repeated Eigen values, to obtain the second equation ~(/) = e" = ()(ol+ ()(, A
di fferentiating the expansion for[( ) on both sides (i.e. Eq I),
1 0 1
!!"[A
dA IO] ,.-, = ()(, =()(oO I:t()(, 0
"
Modern Control Theory 10EE5& Modern Control Theory lOEE55
ao= ex,
o e - 2a, ao - 3a,
Ze" _ e-2t e-/ -e -2r
o b) A -= I 1 ; characteristic equation is AI - A = 0I I - e-I + 2e
as before.
-2 -i-e-' + 2e-" ,-21
ex, = te"
hence OCo = e-' (1 + t) from Eq. (2).
and substituting = 2=-1 I;v:- AI ~
Eigen values are
! 0.
O
A- 2
0
A -I
o
A, = 2.A, = 2.A, = I
=0
c). A =
o characteristic equation is IAI - AI = O. te' = ex, + 4a, (3)
-2 -3
and e'" = e, + ex,A, + a,A;
I A+3=OI
A
2
-I A, = -I,A, = -2.
From Eqs.
= ao + ex, + a,
(2), (3) and (4), solving for ao,a, .,
(4)
d [e-'''x(l)]
ill ",
-Ac' "x(l) te "x(t)
= ~(t -T)Bu{T)dT (2)
(' - /,[.\'9TO - Ax(l)] (3)
hence Eq (2) can be written as Now, th~ state transition matrix ~(l) can be evaluated as
d Ie -"
' x ()]
t = (' -/,
11(1 ) e" ~~(t) = -I lSI - A]-I
ilt
Integrating both sides with time limits 0 and t lilt
1.0128e-a" - 0.0128e- .
.. ~,t-T);,
:1
I.bI28e-u,(H)
- 0.114e-,(H) + 0.114e-.,(H)
- 0.0128e-"('-' 0.114e-a, (I- - 0.114e-,(H)
The Eq (4) is the solution of state equation with input as utt] which represents the change yet) = [! l}x(t); x{O) = I
of state from .1(1,,) to x(1) with input utt). Solution: The
Example 1: Obtain the response of the system for step input of 10 units given state model
as
Dept. of EEE,5J8fT .
Page 67 Dept. ofEEE, SJ8fT
Page 68
Modern Control Theory 10EE55 Modern Control Theory 10EE55
-I
2e -e
-21
e-/ -e -21 Solution: State transition matrix
2~, 2. ~4'
-e --e
3 3
ZIR = ~(t)x(O) I -t . 4 ~"
-e "'t'-e
2e- f _ e-2t e " _ e-2t 1 3 3
-2e~'+2e~2' -e~'+2e~2' -I
x(t) = ~(t)x(O) + ~(t - T )Bu(T)dT
2e-t e" _e" + e-21
_
.'. x(t) = ZlR + ZSR The system has two state variables. XI(t) and X2(t). The control input u(t) effects the state
variable XI(t) while it cannot effect the effect the state variable X2(t). Hence the state
-, variable X2(t) cannot be controlled by the input u(t). Hence the system is uncontrollable.
-e i.e., for nth order, which has 'n' state variables, ifany one state variable is uncontrolled
by the input u(t), the system is said to be UNCONTROLLABLE by input l1(t).
.!.(l+e~2'
2 Definition:
-e ~2' For the linear system given-by
y(t) = Cx(t) = [l l)x(t) X(t) = AX (t) + Bu(t)
y(t) = CX(t) + Du(t)
.!.(l + e~"
= [I I J = 2 is said to be completely state controllable. If there exists an unconstrained input vector
_ e-2t
u(t), which transfers the initial.stntcofthesysrcjn x(to) to its final state X(tf)in finite time
f(tt-to) i.e. fr. It can be seen if' all the initial states arc controllable the system is completely
=.!.(I + e~") - e~2'
2 controllable otherwise the system the system uncontrollable.
is said to completely state controllable iff the rank of controllability matrix Q. is equal to
[1I-1](1I+2J~0
order of system matrix A where the controllability matrix Qc is given by,
ii' + 311+ 2 "0
(II t 2)(iI + I) = () Q, = [B IABIX BI..... yt' sj.
II = -2,11 = - I i.e., if system matrix A is of order nxn then for state controllability
Eigcn vector associated with 11=-2 p(Q,)=n
-2 11- I f
where p Q,.) is.rank of Qc.
1
o 01 v, = I
Example: Using:Kalman's Approach determine the state controllability of the system
described by :
Eigen vector associated with II = - I
y+3y+2y=u+u (I)
I-I + I I 1 with x, = y,x, '= y - u
I 0 - I+2 x, = -3x, :-2x, - 2u
XI =x2 + u
v~ :; 0 f'=
o '. X, =,0 I' x, + I
u(t)
o x, - 2 - 3 x, -2
p.' = _-_1_1 o 0'. I I
A= : ,B =
I -I -2 -3 -2
o
Dept. ofEEE, SJBIT
..
~'lge 71 Dept. of EEE, SjBIT
Page 72
:.
Modern Control Theory lOEE55
Modern Control Theory lOEE55
Verify the result with Gil~_r!,~ upproach.
-2
Now AB = Solution: Kalman's approach: Here,
-2 -3 -2 4
0 0 0 0
-2
Q, = [BIAB]= I A 0 0 1 ,B = o ;AB= 1 ;A 'B = - 3
-2 4
0 -2 -3 I -3 7
o 0 o 2
Z(I) = 0 -I o Z(I) + - I 11(1)
I
o o -2
2
.. u~.,
tt.)
-I -~
.'
Modern Control Theory 10EE55
Modern Control Theory 10EE55
On measurement of yet), we call observe x 1(1) but we cannot observe X2(t)by
measuring alii y(I), hence statc is not completely observable.
ATCT]= 1
-2
Definition: " =[C
00
J\ system described by
x(t) = AX(/) + BU(f) lit =0 pet) < 2
y(/) = CX(/) + P"(/) is said to be completely observable iff with measurement of output Order of system is 2. So, system is unobservable.
yet) over-finite time interval to t If along with input u(t); t to the initial state x(to) c~n be
estimated. Example: Show that the system
Explanation: o. o o
Consider the output equation x(t) = 0 0 1 x(t) + 0 u(t)
y(f) = CX(/) + /)11(1)
-6 -11 -6 1
Assuming /)=0, y(I)=Cx(t).
yet) = [4 5 I!x-(t) is not completely observable by
All the stale variables x(t) can be estimated with measurement of yet) provided C matrix
do not contain the 'zero' clement. If any element of C matrix is zero, the corresponding I) Kalman's approach
slate cannot be estimated and hence system is_un~bscrvablc. 2) Gilbert's approach
Kollman's Approach:
-~------ Solution: Kalman's approach
The system described by the model 0 0 0 0 -6
X(f) = AX(I) + Butt)
,1'(1) = CX(I) T DU(f) A; 0 0 I A = 0 -II
is said to be completely observable iff the rank of observability.matrix ~" is equal to the
-6 -II -6 0 -6
order of system matrix A where,
~,,=[C' A'CT A"C' A" 'C'l 4
For nthorder system, rank of~" must be n. C= [4 5 11 cT 5
i.c. P(~,,) = n
It involves A and C matrices. The pair A. C is observable.
6 -6
Example: Consider the system AT cr : _ 7 ; A r' C T = 5
- 2 () 3
x(t l = 0 _ I x(t) + I lI(f) -I -I
1'(1) = [1 oh(1)
Test the observability using Kalman's approach. 6 -61
t4= "I ~ =0
Solution: Here 5 -7 5
- 1 0 1 -I -I
oj
A= - ;C=jJ
P(~:) <3 add order of system is 3, Therefore system is unobservable.
o .'
-2
o -2 .' Gilltert's approach:
Transferring the system model into observable canonical form with usual
A'e' = procedure ..the output equation will be
() -I 0 o yet) = [0 - 2- 2j;(t
.
+ Bu(t)
'
Let us dssumc that all th~ state variables are measured accurately, and then it is possible to implement
a linear tontrollaw of the form
u(t) = -KX(t)
where,
K = [k,k2 .. .... kn, ] is
(Ixn) feedback gain matrix.
Subs~tute Eq (1) in Eq (I) to obtain closed loop system
model as '
X(t) = AX(I) - BKX(t)
= [A - BK]X(t)
The characteristic Eqn of closed loop system will be
S~-(A-BK) =10
When Eq (4) is evaluated, this yields an nih order polynomial in S containing the n-gains K"
K2, ..... Kn. The control Law design then consists of picking the gains so that the roots of Eq
(4) are in desired locations.
In the next section it can be seen that if the state model is in controllable canonical' form, then the
roots of characteristic Eqn (4) can be chosen arbitrarily. If all the given values of (A-BK) are
placed in len- half of s-plane, the closed loop system is asymptotically stable. And the state Xu)
decays to Zero irrespective of initial state X(O)- the initial perturbation in the state. The system
state can be maintained Zero value in
spite of disturbances that act upon the system. System with this property arc called
REGULATOR SYSTEMS. Control configuration of such system as shown
Let the desired characteristic equation be 6) Determine the desired cha equation as
'S + a,sn-, + , + an
=0 , (16) sn + a,sn-' + a sn-2 + + an = 0
Equating the coefficients of like powers of S from equation ( 15) and (Hi)
7) Determine the elements of gain matrix K, K,. K, ,.... K"
,
}
a, = +Kn
t Kn =a,- as K1 = an - n
K,=43
K2=9
K3=3
If all the state variables are estimated, it is called Full-order state observer. If only few
-'--~---.-.--_j
statevariable are estimated (which can't be measured accurately), the observer is called
Reduced-Qrder state observer.
+ Bu(t)
Design of Full-order observer: In this case all the state variables are to be
estimated. a) Open loop observer: Here it is assumed that all the state variables
.. State model is X(t) = Ax(t)
Y(t)= Cx(t)
(1)
(2)
are estimated
Estimator output is Y(t) = CX(t) (2a)
correctly and initial state are also estimated correctly.
i.e For State model
X(t) = Ax(t) + Bu(t)
where the co-efficients ()(,,()(2 , ....... ()( n are obtained from poly equation
lSI - AI;, 0
8~+ . ,8n-t + ....... + n =0.
.' 4. The observer gain matrix 'm' will be then
m'= [m"m2, ....... mnlT
Example: Consider the state model
o 1 0 o
X(I) = 0 0
1 X(I)+ 0 u
-6 -11 -6 1
Y(I) = [0 0 11X(I)
Dept. of EEE,Sj61T
Page 87 Dept. of EEE, 5j61T
Page 88
Modern Control Theory 10EE55 Modern Control Theory lOEE55
2) Design an observer; where observer-error poles are required to be located at - 2 j3.64 Reduced-order state observer Required when few of
and -5
state variables cannot be measured accurately. Design of Reduced-
Solution: Model in oberserver conamical form will be
order state observer:
o -6 0 In this design procedure, let us assume one measurement Y(t). Which measures one state
X(I) = 0 0 -11 X(I) + 0 X1(t). The output equation is given by
-6 Y(t)=CX(t) (1)
Where C=[ I 0 -----0]
yet) = [0 0 1]X(I) Partitioning the states X(t) into states directly measured X,(t) and states directly non
I) Desired characteristic measurable X,(t), which arc to be estimated. .,.
equation will be .. . _ X1(1)
i.e X(I) - Xe(l) (2)
(S+2+j 3.64) (S+2-j3.64) (S+5)=0.
=> 83 + 982 + 208 + 60 = 0 Partionionly the matrixes accordly & state equation can be written as
hence comparing with 83 + a.s" + a2s + a, =0
X,(t) = a" a" X,(I) + b, u(l) (3)
Q] =9, Q, =20, QJ =60 (1)
X. (I) a., ae X,(I) b,
2) Let the observer be m = [m] , m, , m, l'
o 0 -6 0 0 rn, and Y(I) = [1 Or x (I) (4)
X,(I)
[A-mc]= 1 0 0 m2
from equation (3)
-'6' 0 m3
X. (t) = a.eXe(I),+ a ;(I) +J'J (5)
o 0 -(6+m,) Knowninpul
(same as X(I) = AX(I) + bu(t) + m(y(l)- y(I)) where yet) = a,.X. (I) Solution: :
Detining Xe (I) = X. - Xe From output Equation it follows that states X,(t) & X2(t) cannot
be measured as corresponding elements ofC matrix are zero. Let
Xe (I) = x, (1).- Xe (I) us write the state equation as
r Solution: S'ystem is observable & Controllable. Using the procedure of controller design
& observer design, the controller will be
K=[4 I I] and observer will be
1
M=[12 25 _72]T
I
I leot'lt",oItf:"T bb5f.T'it y
Transfer function of uncompensated system will be
L _ ;<-CC"1r""""l"': Mu(S)=qSI-Ar'B.
- - - --00\ .
M _ 1
'tSI - S3 + 6S 2 + 11S + 6
Transfer function of compensator: Transfer function of compensated system will be
Now X(t) = AX(I) + Bu(t) + m(y - eX(I)) (I) Me(s)= U(S) = K[SI- (A - BK - mCW'm
YeS)
'Prop')J1ional- Derivative (PO) controllers and Proportional - fntegral- Derivative (PID) controllers. Mathematically the integral controller mode is expressed as pet)
The block diagram of a basic control system with controller is shown in Figure. The error detector compares The constant K. 'is called integral constant. The output tram the controller at any instant is the area
the feedback signal b(l) with the' reference input nt) to generate an error. crt) = ret) _ bet).
under the actuating error curve up to that instant. If the error is zero, the controller output will not
change. The inte,grar controller is relatively slow controller. It changes its output at a rate which is
Controller output -:
(p) dependent on the integrating time constant, until the error signal is cancelled. Compared to
r(l) 0(\) j the proportional controller, the integral control requires time to build up an appreciable output.
Controller
~)- However it continues to act till the error signal disappears. Hence, with the integral controller the
Error steady state error can be made to zero. The reciprocal of integral constant is known as integral time
.. detector
constant T;. j.e., T; =. ilK;.
L.._~ -j Feedback element 1------'
/
Feedback Derivative Controller:
Signal
In this mode, fhe output of the controller depends on the rate of change of error with respect to time.
Proportional Controller:
In the proportional control mode, the output of the controller is proportional to the error e(t).
The relation between the error and the controller output is determined by a constant called proportional
Hence it is alsb known as rate action mode or anticipatory action mode.
The mathemafical equation for derivative controller is p(t)
' .
_ K. ~
- + P"
Where Kais the derivative gain constant. The derivative gain constant indicates by how much
percentage the controller output must change for every percentage per second rate of change of the
gain constant denoted as K. i.e. p(t) = Kp eu). error. The advantage of the derivative control action is that it responds to the rate of change of
error and can produce the significant correction before the magnitude of the actuating error
Though therc, exists linear relation between controller output and the error, for zero error the becomes too large. Derivative control thus anticipates the actuating error, initiates an early corrective
controller output should not be zero as this will lead to zero input to the system or process. Hence, action and tends to increase stability of the system by improving the transient response. When
there exists some controller output Po for the zero enor. Therefore mathematically the proportional the error is zero or constant, the derivative controller output is zero. Hence it is never used alone.
PI Controller:
Dept. of EEE. 5J8fT
Page 95
Dept. of EEE, SJBIT
Page 96
..,
Modern Control Theory lOEESS Modern Control Theory lOEES5
This is a composite control mode obtained by combining the proportional mode and
integral mode. The mathematical expression for PI controller is
The transfer function is given by g~ I . It mcreases the damping ratio and reduces overshoot.
2. It r~duc~s the rise time and makes response fast.
3. It r;eduees the settling time
4. The type of the system remains unchanged.
5. Steady state error remains unchanged.
The advantage of this controller is that the one to one correspondence of In general it improves transient part without affecting steady state.
proportional controller and the elimination of steady state error due' to integral
The transfer function is given by
controller. Basically integral controller is a low-pass circuit.
The PI Controller has following effects on the
system.
PD Controller: This is to note that derivative control is effective in the transient part .~' the
This is a composite control mode obtained by combining the proportional
response as error-is varying, whereas in the steady state, usually if 'any error is there, it is
mode and derivative mode. The mathematical expression for PI controller is
constant and does not vary with time. In this aspect, derivative control is not effective in
, deft) .
P(f'J T(p,(t, 4- IC,~ -e- P"
-the--stearr-y-state part of the response, In the steatly state part, if any error is there,
The transfer function is given by Kt K,s integral control will be effective to give proper correction to minimize the steady state
eITOr. An integral controller is basically a low pass circuit and hence will not be effective in
transient part of the response where error is fast changing, Hence lor the whole range of
time response both derivative and integral control actions should be provided in addition
to the inbuilt proportional control action for negative feedback control systems,
PID Controller:
This is a composite control mode obtained by combining the proportional mode,
integral mode and derivative mode. The mathematical expression for PI controller is
UNlT- 6
Non-linear systems: Introduction, behavior of non-linear system, common physical non linearity-
saturation, friction, backlash, dead zone, relay, multi variable non-linearity.
3 Hours
Many practical systems are sufficiently nonlinear so that the important features of
Figure 6.1
their performance may be completely overlooked if they are analyzed and designed
A nonlinear system, when excited by a sinusoidal input, may generate several
through linear techniques. The mathematical models of the nonlinear systems are
harmonics in a~dhion to the fundamental corresponding to the input frequency. The
represented by nonlinear differential equations. Hence, there are no ~en'eral methods
amplitude' of the :fundamental is usually the largest, but the harmonics may be of
for the analysis and synthesis of nonlinear control systems. The f;~tjhat
significant implitude in many situations.
superposition principle does not apply to nonlinear systems makes generalisation
difficult and study of many nonlinear systems has to be specific for typical situations.
Another peculiar characteristic exhibited by nonlinear systems is called
jump phenomenon. For example, let us consider the frequency response curve of
6.2 Behaviour of Nonlinear Systems
spring-mass- damper system. The frequency responses of the system with a linear
The most important feature of nonlinear systems is that nonlinear systems do not
spring, hard spring and soft spring are as shown in Fig. 6.2(a), Fig. 6.2(b) and Fig.
obey the principle of superposition. Due to this reason, in contrast to the linear case,
6.2(c) respectively. For a hard spring, as the input frequency is gradually
the response of nonlinear systems to a particular test signal is no guide to their
increased from zero, the measured response follows the curve through the A, Band
behaviour to other inputs. The nonlinear system response may be highly sensitive to
C, but at C an increment in frequency results in discontinuous jump down to the point D,
input amplitude. For example, a nonlinear system giving best response for a certain
after which with further increase in frequency, the response curve follows through DE.
step input may exhibit highly unsatisfactory behaviour when the input amplitude
If the frequency is now decreased, the response follows the curve EDF with a jump up
is changer' Hence, in a nonlinear system, the stability is very much dependent
to B frorq the point F and then the response curve moves towards A. This
on the input and also the initial state.
phenomenon which is peculiar to nonlinear systems is known as jump resonance. For
Further, the nonlinear systems may exhibit limit cycles which are self-
a soft spring, jump phenomenon will happen as shown in fig. 6.2(c).
sustained oscillati~ps of fixed frequency and amplitude. Once (he system trajectories
converge to a limit cycle, it will continue to remain in the closed trajectory in the state
space identified as limit cycles. In many systems the limit cycles are undesirable
particularly when the amplitude is not small and result in some unwanted phenoraena. '.
,
,
Modern Control Theory 10EE55 Modern Control Theory 10EE55
When excited by a sinusoidal input of constant frequency and the amplitude is increased ~ Liapunov's Method for Stability: The analytical solution of a nonlinear system is rarely possible.
low values, the output frequency at some-point exactly matches with the input frequency If a numerical solution is attempted, the question of stability behaviour can: not be fully answered
and continue to remain as such thereafter. This phenomenon which results in .a as solutions to an infinite set of initial conditions are needed. The Russian mathematician A.M.
synchronisation or matching of the output frequency with the input frequency is called Liapunov introduced and formalised a method which allows one to conclude about the stability
frequency entrainment or synchronisation. without solving the system equations.
However, starting with the classical techniques for the solution of standard nonlinear nonlinearities. The nonlinearities which are present in the components used in system due
differential equations, several techniques have been evolved which suit different types of to the inherent imperfections or properties of the system are known as inherent
analysis. It should be emphasised that very often the conclusions arrived at will be useful for nonlinearities. Examples are saturation in magnetic circuits, dead zone, back lash in
the system under specified conditions and do not always lead to generalisations. The gears etc. However in some cases introduction of nonlinearity may improve the
commonly used methods are listed below. performance of the system, make the system more economical consuming less space and
more reliable than the linear system designed to achieve the same objective. Such
Linearization Techniques: In reality all systems are nonlinear and linear systems areonly nonlinearities introduced intentionally to improve the system performance are known as
approximations of the nonlinear systems. In some cases, the linearization yields useful intentional nonlinearities. Examples are different types of relays which arc very
information whereas in some other cases, linearised model has to be modified when the frequently used to perform various tasks. But it should be noted that the improvement
operating point moves from one to another. Many techniques like perturbation method, series in system performance due to nonlinearity is possible only under specific operating
approximation techniques, quasi-linearization techniques etc. are used for linearise a nonlinear conditions.r other conditions, generally nonlinearity degrades the performance of the
system. system.
Phase Plane Analysis: This method is applicable to second order linear or nonlinear systems 6.5 Common Physical Nonlinearitles: The common examples of physical
for the study of the nature of phase trajectories near the equilibrium points. The system nonlinearities are saturation, dead zone, coulomb friction, stiction, backlash, different
behaviour is qualitatively analysed along with design of system parameters so as to get the types of springs, different types of relays etc.
Describing Function Analysis: This method is based on the principle of harmonic phenomenon in magnetic circuits and amplifiers.
linearization in which for certain class of nonlinear systems with low pass cha,racteristic. This
method is useful for the study of existence of limit cycles and determination of the amplitude, Dead zone: Some systems do not respond to very small input signals. For a particular
frequency and stability of these limit cycles. Accuracy is better for higher order systems as they range of :ippqt, the output is zero. This is called dead zone existing in a system. The
have better low pass characteristic. input-output curve is shown in figure.
Dept. of EEE, SJBIT Page 101 Dept. of EEE, SJBIT Page 102
" .
Modern Control Theory lOEE55
Modern Control Theory lOEE55
needed to close the relay than the current at which the relay drops out, the characteristic always
exhibits' hysteresis.
Saturation Deed-zone
Figure 6.3 Multivariable Nonlinearity: Some nonlinearities such as the torque-speed characteristics of a
servomotor, transistor characteristics etc., are functions of more than one variable. Such
nonlinearities are called multivariable nonlinearities.
. Figure 6.4
Relay: A relay is a nonlinear power amplifier which can provide large power
amplification inexpensively and is therefore deliberately introduced in control systems. A relay
controlled system can be switched abruptly between several discrete states which are usually oft;
full forward and full reverse. Relay controlled systems find wide applications in the
control field. The characteristic of an ideal relay is as shown in figure. In practice a relay has a
definite am~t of dead zone as shown. This dead zone is caused by the facts that relay coil
requires a finite amount of current to actuate the relay. Further, since a larger coil current is
"
"
Modern Control Theory 10EE55
Modern Control Theory lOEE55
variables are zero. These points arc called singular points. These are in faet equilibrium points
UNIT -7 PHASE PLANE ANALYSIS of the system. If the system is placed at such a point. it will continue to lie there if left
UNIT-7 undisturbed. A family of phase trajectories starting from different initial states is called a
Phase plane method, singular points, stability of nonlinear system, limit cycles, constructlon of phase phase portrait. As time t increases, the phase portrait graphically shows how the system moves
trajectories.
in the entire state plahe from the initial states in the different regions. Since the solutions from
.
,7 Hours
each of the initial conditions are unique, the phase trajectories do not cross one another. If the
system has non linea; elements which are piece-wise linear, the complete state Jjlace can be
Phase plane analysis is one of the earliest techniques developed for .the stugy of second
divided into different regions and phase plane trajectories constructed for each of the regions
order nonlinear system. It may be noted that in the state space formulation, the siate variables
separatel y.
chosen are usually the output and its derivatives. The phase plane is thus'a state plane where the
two state variables XI and xz are analysed which may be the output va~iable y and its 7.3 Phase Plane
, Method
Consider the homogenous second order system with diffcrcnnai equations
derivative y. The method was first introduced by Poincare, a French mathematician. The
dx
method is used for obtaining graphically a solution of the following two simultane~us equations
of an autonomous system.
This equation may be written in the standard form
....fl(X1'X~;
~f,Cx"x,) where and narc t4e damping factor and undamped natural frequency of the system. Defining the state
Where it f,~X;;.f,) and .t, {,('X"x,) are either linear or nonlinear functions of the state variables as x = x 1 and :\-= X2, we get the state equation in the state
variables XI and x2 respectively. The state plane with coordinate axes x I and X2. is called the variable form as ,
phase plain. In many cases, particularly in the phase variable representation. of systems, take the "
form
- ."t~
These equations may then be solved for phase variables x I and x2. The time response plots of
The plot of the state trajectories or phase trajectories of above said equation thus gives an
x I, X2 for various values of damping with initial conditions can be plotted. When the differential
idea of the solution of the state as time t evolves without explicitly solving for the state. The
equations describing the dynamics of the system are n~lj*ar. it is in general not possible to
phase plane analysis is particularly suited to second order nonlinear systems witli no input or u
obtain a closed form solution of x I, x2. For example, if the spring force is nonlinear say (k ] x +
constant inputs. It can be extended to cover other inputs as well such as ramp inputs, pulse
inputs and impulse inputs. k2x3) the state equation takes the form
Dept. of EEE, SJBIT Page 105 Dept. of EEE, SJBIT Page 106
"
"
Modern Control Theory '. 1(1.EE55 Modern Control Theory lOEE55
with axes that correspond to the dependent variable x I and X2 is callcd'phase-plane. The curve
described by the state point (x ),X2) in the phase-plane with respect to time is called a phase
Therefore, the locus of constant slope ofthe trajectory is given by I;(xlox,) ~ Mfl(xl,x,)
The above equation gives the equation to the family of isoclines. For different values of
M, the slope of the trajectory, different isoclines can be drawn in the phase pIane.
Knowing the value of M on a given isoclines, it is easy to draw line segments on each of these
isoclines. . Phase plot
Consider a ~mple linear system with state equations
- x., Figure 7.1
= -~y')- XI
The Procedure for construction of the phase trajectories can be summarised as
below:
Dividing the above equations we get the slope of the state trajectory in the XI-X2plane as
d.1z _-x,-x'_M I. For the given nonlinear differential equation, define the state variables as XI and X2
and obtain the state equations as
;i;. '" x,
2. Detenninc the equation to the isoclines as
For a constant value of this slope say M. we get a set of equations 'dxz '" {(xl'x:) = M
-1
dx, r,
- ... ,_x 3. For typical values ofM, draw a large number of isoclines in XI-X2
(M.;, 1) J
plane
which is a straighi line in the xl-x2 plane. We can draw different lines in the xI-x2 pIane for 4. On each of the isoclines, draw small line segments with a
slope M.
different values of M; called isoclines. If draw sufficiently large number of isoclines to cover the
5. From an initial condition point, draw a trajectory following the line
complete state space as shown, we can see how the state trajectories arc moving in the state segments with slopes M on each of the isoclines.
plane. Different trajectories can be drawn from different initial conditions. A large number of Example 7.1: For the system having the transfer function .'.~,) and a relay with
such trajectories together form a phase portrait. A few typical trajectories arc shown in figure dead zone as nonlinear element, draw the phase trajectory originating from the initial
condition (3,0).
given below.
Dept. ofEEE. SJBIT Page 107 Dept ofEEE, SJBIT Page 108
Modern Control Theory 10EE55 Modern Control Theory 10EE55
--r
.i...:l;.~v..~
Here u = -I so that on substitution we get x, =- -.\':+1 or r..-~ 1
1..::!'.J..~ ._.~~ u
These are also lines parallel to x - axis at a constant distance decided by the value
of the slope of the trajectory M.
The differential equation for the system is ,,+; -" where u is
given by M 0 113 I 2 3 -5 -4 -3 -2 I
I ,
I
11 -
-!-1 II.>
0 If-l"' ..",
f I 0.75 0.5 113 0.25 -0.25 I -0.33 -0.5
~_~~J
{ -1
tIe'" -1
-t-X,
Since the input is zero, e = r - c = -c and the differential equation in terms of error Region n Region I
will be -11 Region III
;;..f..i- s=o
113
Defining the state variables as X, = e and X2 = ;. we get the state I
equations as 3
.' -5
.' -3
-713
The slope of the trajectory is -2
given by
dA.~ -X ...-JI
dx; - --;;- - 1.1
Figure 7.2: Phase portrait of
Equation to the isoclines is Example I
given by
-x ..-u Isoclines drawn for all three regions arc as shown in figure. It is seen that
.:.t~-~
trajectories from either region I or 2 approach the boundary between the regions
We can identify three regions in the state plane depending on the values of e and approach the origin along a line at -I slope. The state can settle at any value of
=x., x, between -I and + I as this being a dead zone and no further movement to the
origin along the x.-axis will be possible. This will result in a steady state error.
Re ,,= x. ~ 1 the maximum value of which is equal to half the dead zone. However, the
gio
presence of dead zone can reduce the oscillations and result ill u faster response. It,
n
order that the steady state error in the output is reduced, it is better to have as small "
1: dead zone as possible.
Here u = I,so that the isoclines are given by x~ = -.~- J or 1:2 = ,,-. rc
Example 7.2: For the....s.ystcmhaving a closed .1001' transfer function
For different values of M, these are a number of straight lines parallel to the
plot the phase trajectory originating from the initial condition (-1,0).
x-axis.
..
Modern Control Theory lOEE55 Modern Control Theory lOEE55
The slope of the trajectory
is given by
dx~ _ -2x~ - 5'''"1 ~ If ..,.A,1
dX1 ...",
-SXt 10
.".1,,'" -;;-::;:-;+ ,~.,_;__?
When~Xl= 0,
"
-Sx, 10 "
-;:;;-:;; 4- :v-:;:; - 0 or X t - 1
_ c: 111
Figure 7.3
X, -~Xl';"~ The isoclines are drawn as shown in figure. The starting point of the trajectory is marked at (_
1,0). At (-1,0), the slope is co, ie., the trajectory will start at an angle 90. From the
next isoclines, the average slope is (8+4)/2 = 6, ie., a line segment with a slope 6 is drawn (at an
-c: in angle 80.5).The same procedure is repeated and the complete phase trajectory will be obtained
x' - W'\j -e- 1ii as shown in figure.
7.3.2 Delta Method:
When M = -2', When M = -4, The delta method of constructing phase trajectories is applied to systems of the form
:C. - (C " ..;.j(x.x.tJ-~
2 Where fL".",t") , may be linear or nonlinear and may even be time varying but must be
continuous and single valued.
S I,'
"\'-iA,-- With the help of this method, phase trajectory for any system with step or ramp or any time
"
"". varying isput can' bc conveniently drawn. The method results in considerable time saving when
When M = -6, When M = -10, a single or a few phase trajectories are required rather than a complete phase portrait.
<; III While appl~n~ the delta method, the above equation is first converted to the form
x: ='4""-4 .'i -+ Cl~[X +
In general, 8(x,f.~f'd~tJJds llpon the variables x, ,. and t, but for short intervals the
<; I,' changes in th,se variables are negligible. Thus over a short interval, we have
x: =fix,-R' fiJ i' ~S(Q' +is a eolil;tant.
Let us choose the state variables as x, = X; ~, - x/lAJ", then
~TL =~X2
12 - -W'/~l + e;
Dept. of EEE. SJBIT
Page 111 Dept of EEE, SJBIT
Page 112
..
Modern Control Theory lOEESS Modern Control Theory IOEE55
Xi -Xl which describes physical situations in many nonlinear systems. In terms of the Slatevariables
""\~~
- -X., - x; A, - T c,ncf .:t_: - i., we obtain
The above equation can be rearranged as
- x~
""
x:--rx,-!-x .;....~-X,)
So that ,~- x, +."r-x, The figure shows the phase trajectories of the system for > 0 and < 0. In case of > 0 we
At initial point is calculated as = 0+ I- I = O. Therefore, the initial arc is centred at observe that for large values of x,(O), the system response is damped and the amplitude or
point (0, 0). The mean value of the coordinates of the two ends of the arc is used to x,(t) decreases till the system state enters the limit cycle as shown by the outer trajectory. On the
calculate the next value of and the procedure is continued. By constructing the small other hand, if initially XI(O)is small, the damping is negative. and hence the amplitude of x,(I)
arcs in this way the complete trajectory will be obtained as shown in figure. increases till the system state enters the limit cycle as shown by the inner trajectory, When < 0,
the trajectories moves in the opposite directions as shown in figure.
Unstable
limit
cycle
Figure 7.5
A limit cycle is called stable if trajectories near the limit cycle, originating from outside
Figure 7.4 or inside, converge to that limit cycle. In this case, the system exhibits a sustained
Dept. of EEE. SJBIT Page 113 Dept. of EEE. SJBIT Page 114
:
Modern Control Theory lOEE55 Modern Control Theory lOEE55
oscillation with constant amplitude. This is sliown in figure (i). The inside of the limit cycle is
an unstable region' in the sense that trajectories diverge to the limit cycle, and the outside is a Where, k, = ( 2/ i) 0 so that the trajectories
stable region in the sense that trajectories converge to the limit cycle. become a set of parabola as shown in figure (b) and
the equilibrium point is called a node. In the
A limit cycle is called an unstable one if trajectories near it diverge from this limit cycle. In this
case, an unstable region surrounds a stable region. If a trajectory starts within the stable region, it original system of coordinates, these trajectories
appear to be skewed as shown in figure (c).
converges to a singular point within the limit cycle. If a trajectory starts in the unstable region. it
diverges with time to infinity as shown in figure (ii). The inside of an unstable limit cycle is the If the eigen values are both positive, the nature of
stable region. and the outside the unstable region. \ -( \ \ the trajectories does not change, except that the
7.5 Analysis and ClassifIcation of Singular Points: ~ . J\\i 7',
>~. trajectories diverge out from the equilibrium point
as both ZI(t) and Z2(t) are increasing exponentially.
Singular points are points in the slate plane where x', f: .,rAt these points the slope of the \ \ ,," The phase trajectories in the XI-X] plane are as
trajectory dx)/dx, is indeterminate. These points can also be the equilibrium points of the ('1'........ ""....... shown in figure (d). This type of singularity is
nonlinear system depending whether the state trajectories can reach these or not. Consider a ;".._, - identified as a node, but it is an unstable node as the
lincariscd second order system represented by . r trajectories diverge from the equi!ibrium point.
-
Using linear transformation x = Mz, the equation can be transformed to canonical fonn
:
;
i
(.,""...........---
"
Saddle Point:
Consider now a system with eigen values are real.
distinct one positive and one negative. Here, one of
the states corresponding to the negative eigen value
converges and the one corresponding to positive eigen value diverges so that the
trajectories are given by Z2 = C(ZI)'kor (Z,)kZ2 = c which is an equation to a rectangular
Where, ,and ,are the roots of the characteristic equation of the s~. hyperbola for positive values of k. The location of the eigen values, the phase portrait in
The transformation given simply transforms the coordinate axes from X'-X2 plane to ZI-Z2plane ZI - Z2 plane and in the x, - X2 plane are as shown in figure. The equilibrium point
having the same origin, but docs not affect thc nature of the roots of the characteristic around which the trajectories are of this type is called a saddle point.
equation. Thc phase trajectories obtained by using this transformed state equatjon still carry the
same information except that the trajectories may be skewed or stretched along the coordinate
axes.In general, the new coordinate axes will not be rectangular. jrll
The solution to the state equation being given by
z,lt,- ,,\"z.,(U'
The slope of the trajectory in the z, - z, plane is given by
1:""
(a) Location of cigcnl'l1uc.1
XI
.'
Based on the nature of these cigcn values and the trajectory in z, - z, plane, tho sihgular
points arc classified as follows.
Nodal Point:
(e)Saddlepointin ("" xv-plane
Consider cigen values arc real, distinct and negative as shown in figure (a). For this case
the equation of the phase trajectory follows as =:
= =J" .Phase trajectories around saddle point
Integrating the above equation, we get .Y,~ +y,; = 1<2 which is an equation to,a
The slope circle of radius R. The radius R can be evaluated from the initial conditions. The
trajectories are thus concentric circles in YI-Y2 plane and ellipses in the XI-X,
dyz -cuYI + aX2 _ Yz - kYl plane as shown in figure. Such a singular points, around which the state
dy, - a)l, + wy, )', - k"z trajectories are concentric circles or ellipses, are called a centre or vortex,
.,,
-o.-jtA
Let the state variables be x, - v ami
The corresponding state model is
-"!: .i .
..\';1 - x ..
(b) Ph.. u.nw:rmes!UDl>le b.."lr.I
.
,
-/<0
The characteristic equation is I-A =0
. II.2
i.e,
-I
;.+:.1
- n or
(b)Ptluetmjec:tnfieltntable incus
Ph.le tr"je<.t.o1Jes
.round unsteble focus I, 2 = -2, -I. Since the roots are real and negative, the type or singular point is stable
node.
Dept. of EEE, SJBIT Page 117 Dept. ofEEE, SJBIT Page 118
Modern Control Theory lOEESS
Modern Control Theory
..
Example 7.S: For the nonlinear system having differential equation;
. lOEESS
..
.'
Example 7_~:
Determine tlie kind of singularity for the following differential equation.
y- (n.l
, ,
_.!j,2)y +]. _),2 ~ O. find all singularities.
r.;- O.5:i: + 2.- + r2 =0
Defining the state variables as l' = X " 'I' = x" the state equations are
Let the statevariables be x 1 = x ulu.t x, = i-
The,correspondihg state model is
x, := x 2 :
i :e; .:':l=Ol",'-I I So tfiat rhe singuler points y, -On ...d -2"",-,,,; i....,x,(x,.,I.2) _ 0
at'df. a:
d!'j - [-(2 ,~ ,
The Jacobean matrix J =
11 1 -'lox~l The Jacobean matrix J =
.;};"a:;- (]
I
l2 -1
..;+ 0"
1= r
i.e., ( -0.1)+ I = 0 or '- 0.1 + I = 0
i.e., ( + 0.5) + 2 = 0 or 2 + 0.5 + 2 = 0
AHA, - O.S( 11 1 + jv'J.99;
A,.A. - (-O.2~ =i Jl.:i',J'
The cigen values arc complex with positive real part. The singular point is an
unstable focus. The eigen values are complex with negative real parts. The singular point is a
stable focus.
Linearization around (-1,0)
J(-I.O; - r~ 11
1 1
II
Linearization around (-2 0)
I( -7..0; - [~
1 .
_flO;
The characteristic equation will be
The characteristic equation will be
I )1 -'1
-\-01
1- n
1 -1 I
Therefore 2 _ 0.1 _ I = 0 1 -1 ,(+0." - n
Therefore 2 + 0.5 _2 = 0
I. 2 = 1.05 and -0.98. Since the roots arc real and one negative and another
positive, the singular point is a saddle point. I, 2 = 1.19.'and -1.69. Since the roots are real and one negative and another positive.
the singular point is a saddle point.
Dept. of EEE, SjBIT
Page 119 Dept. of EEE. SI BIT
Page 120
-
.'
6 Hours
For linear time invariant (LTI) systems, the concept of stability is simple and can be
formalised as per the following two notions:
a) A system is stable with zero input and arbitrary initial conditions if the resulting trajectory
tends towards the equilibrium state.
b) A system is stable if with bounded input, the system output is bounded.
Fig. 8.1 Global and local stability
Dept. of EEE, SJBIT Page 121 Dept. of EEE, SJBIT Page 122
Mo~_ern Control Theory lOEE55 Modern Control Theory 10EE55
The above said definiuons are represented graphically in Fig. 8.2
A scalar function Vex) is said to be negative semi-definite in a region if it is negative for all
states in the region: and except at the origin and at certain other states, where it is zero.
A scalar function Vex) is said to be indefinite if in the region it assumes both positive and
negative values, no matter how small the region is.
Examples:
a., au/
If....,
:: ~. 0.......... rl1aT' au
a" lit"
a 1 "1 ;...0
of the system. The most popular among this is ealled the "Second Method of Liapunov" or
0"
a"l (I., Sl......
"Direct Method of'Liapunov". ' This method is very general in its A necessary and sum bent condition in order that the quadratic form xTAx, where A is an nxn real
formulation and can be used 10 study of stability of linear or nonlinear systems. The method symmetric matrix,. be negative definite is that the determinant of A be positive if n is even and
is called 'direct' method as it docs nut involve the solution of tile system diffct1lbti~1 negative if n is odd, and the successive principal minors of even order be positive and the
equations and stability information is available without solving the equations which is successive principal minors of odd order be negative.
definitely an advantage for nonlinear systems. The stability information obtained by
this method is precise and involved no approximation.
First .Met~d of Liapunov: . The first method of Liapunov, though rarely talked about, is e
essentially 3 theorem stating the conditions under which system stability information can be
inferred by examining the simplified equations obtained through local linearization. This
theorem is applicable only to autonomous systems. n
8.4 Sign Definiteness
: il-1"
Let V(x" Xl, Xl, Xn) be a scalar function of the state variables x, Xl, Xl, , x., Then
the following definitions arc useful for the discussion of Liapunov's second method. : a~. -c 0.......... IAI '" 0 :fn If n
~ a'Q
8.4.1 Scalar Functions:
A scalar function V(x) is said to be positive definite in a region A necessary and sufficient condition in order that the quadratic form XT Ax, where A is an nxn real
if Vex) > 0 for all nonzero symmetric matrix, be positive semi-definite is that the determinant of A be singular and the
states x in the region and V(O)~ O.
successive principal minors of the determinant of A be nonnegative.
A scalar function Vex) is said to be negative definite in a region if Vex) < 0 for all nonzero i.e.,
states x in the region and V(O)= O.
A scalar function V(x) is said to be positive semi-definite in a region if it is positive for all all
states in the region and except at the origin and at certain other states, where it is zero. a., l
:> 0..U-21
n~! a.~ Then the equilibrium state at the origin is uniformly asymptotically stable. If in addition,
V(x,1) .....00 a-s Uxl! .....00 then the equilibrium state at the origin is uniformly
Example 8.1: Using Sylvester's criteria, determine the sign definiteness of the asymptotically stable in the large.
following quadratic forms A visual analogy may be obtained by considering the surface ~'= ~x; + ~kxi'- This is a
lOxi + x# + + 4-"1X, + 6r,x. + 1x.x, cup shaped surface as shown. The constant V loci are ellipses on the surface of the cup.
Let (,,~, ,g) be the li1ITiiit1:ondition. If one plots trajectory on the surface shown, the
Successive
2l-Y~]
1 1
1 .j.
t IAI:> n
principal
point of the cup which is the equilibrium point.
..
representative point X(I) crosses the constant V curves and moves towards the lowest
A _
-1
[-t
1 _< -1] 1."",- n
-2 ~1~
Successive principal minors are
-I
-3
11 ", n
2. ~!(XJ' t.~ is ut!'!i:allvt! GQmid"nn.ih~ Let us choose another V(,,) = xl+ 2x;
1-~(0Ct,xc.to).( does not vanish identically in t to lor any to and any xn Then Vex) -' 4x1.1:1+ h,;i,- 4).';'(.(,1+
4X,( -,Y, - x~) -
0, wheJ0(t. X, . tc'; denotes the trajectory or solution starting from XII at tu.
This is negative semi definite function. lfV(,,) is to be vanish identically for t to, then
Then the equilibrium state at the origin of the system is uniformly asymptotically stable in X2 must be zero for all t 10. :. i., - II. Tlu",,i, - -x, - .~: - !l !.~." 1must l!~zero
the large.
This means that V(;~) vanishes identically only at the origin. Hence the equilibrium state
If however, there exists a positive definite scalar function V(x,t) such that V(~.[) is at the origin is asymptotically stable in the large.
identically zero, then the system can remain in a limit cycle. The cquilibriqw;!state at the
origin, in this case, is said to be stable in the sense of Liapunov. Example 8.7: Consider a nonlinear system governed by the state equations
Theorem 3: Suppose that a system is described by x = f(x,t: where j(O,t) = 0 lor all t = -x,
to. If there exists a scalar function W(.r,t) having continuous first partial derivatives and
satisfying the following conditions Let us choose ~'(x) = xr+xi
I. IV(x.1) is positive definite in some region about the origin.
Then vex) =' 2x,x, + Zx,:i";, 'l'\l( -x, +
2. ~-V:"',t; is positive definite in the same region, then the equilibrium stateat
the origin is unstable. = -2XI -I: 1,..1x: - J"'i = -:lxf( 1- 2X1XZ)
'"
Example M.4: Determine the stability of following system using Liapunov's method.
Therefore, for asymptotic stability we require that the above condition is satisfied. The
Let us choose ~'(x) = :'f~"" xi region of state space where this condition is not satisfied is possibly the region of
instability. Let us concentrate on the region of state J space wIn:fe this condition is
satisfied. The limiting condition for such a region is.The dividing lines lie in the first and
third quadrants and are rectangular hyperbolas as shown in Figure. 8.4. In the second and
This is negative definite. Hence the system is asymptotically stable.
Example 8.5: Determine the stability of following system using Liapundv'a method.
fourth quadrants, the .
. inequality is satisfied for all values ofx I
region of stabiJity and' possible instability.
and X2. Figure 8.4 shows the
Example 8.6: Determine: tI\"c stability of following system using Liapunov's method.
"
S,abJequad,.,,,
Let liS choose V(X) - x7 ..;...r.~
-xT!
This is an indefinite function.
Dept. of EEE, SJBIT Page 127 Dept of EEE. Sj BIT Page 128
'.
Modern eory
this equilibrium state, it does not necessarily mean that the motions are unstable x'rx
outside the region . ATp-j. pj. --1
3. For a stable or asymptotically stable equilibrium state, a V function with the
required properties always exists. [
I'll l'11.
[ F,
t:
1'121 o
P..1
[ nI -1.11'' 1'12
-I ~" -I
8.6 Stability Analysis of Linear Systems f... -2P., P." -~, - P,,] ~ [-1
Theorem: The equilibrium state x = 0 of the system given by equation 1: = A x is
LFI! - P':Ji- ?P" 21'" n
asymptotically stable if and only if given any positive definite Hermitian matrix Q (or -2P'2 =-1
positive definite real symmetric matrix), there exists a positive definite Hermitian matrix P
(or positive definite real symmetric matrix) such that A'P + PA = -Q. The scalar P" - PI2 - P22 = 0
function x:' Px is a Liapunov function for the system. 2PI2-2Pn=-1
V(,t} = "'p,, Solving the above equations,
v(~l = x'Px + k'PX = (A.,\}Px";' x'PAx = ~.A' P::: + ,,'FAx PII = 1.5; PI2 ~ 0.5; and
P22 = I
where Q = (A'P + PAl
:. P _ [1.~ 0.5]
O.~ 1.0
Hence, for the asymptotic stability of the system of .f - A", it is sufticient that Q be 1.5 > 0 and det(P) > O. Therefore. P is positive define. Hence, the equilibrium
positive definite. Instead of first specifying a positive definite matrix P and examining state at origin is asymptotically stable in the large.
whether or not Q is positive definite, it is convenient to specify a positive definite matrix
Q first and then examine whether or not P determined from ,.q.1I'is -elso positive The Liapunoy function Vex) =
definite. Note that P being positive definite is a necessary condition. . x'rx
:
0.5' [X(
1.0. Y,.
.tilW::
I. If tI'(~) -x'Qx does not vanish identically along any trajectory, then:Q may = o.S(3xf +'2Xl:C~ + 2x;.)
be chosen to be positive semi definite. . , l'(x! = nS[6., ", + 71',( -.r, - ":) + lx,,,,, +- 4>::t.+x I
2. In determining whether or not there exists a positive definite Hermitian or
real symmetric matrix P, it is convenient to choose Q = I, where I is the identity = -(:::f.+x;)
matrix. Then the elements of P are determined ,fJ'<Jml. JU - _I and the matrix P is
tested for positive definiteness. Jl:xam'ple8.9: Determine the stability of the equilibrium state of the following system.
X]
. [~ ":' -1+}
r-?
.
A'P+PfJ- -T
-? il
.
.- I';'
r -l-j -3
OJ
-{
O.:'I7~ -O.1:!~ - }u.n~l This choice is permissible since V-:x.l - -x'Qxcannot be identically equal to zero
.' p-
r-0125 + 10.1.25 (t.25
except at the origin. To verify this, note that
~'(,,) being identically zero implies that X3 is identically zero. If X3 is identically zero,
then x, must be must be zero since we have 0 = - kx, - O. If XI is identically zero, then X2
Example 8.,10: Determine the stability of the system described by
must also be identically zero since 0 = X2. Thus Vex) is identically zero only at the
[~Jl- r-1'
x -: I
-'J'
-J..
[;'(1"
X,.
origin. Hence we may use the Q matrix defined by a psd matrix.
Let us solve,
Assume it Liapunov function V(x) = XTpX
;;1'0 -+ .0.< - =I
-~n -k] ~"
0 ?Zl
n
o
IT 1 -1 .tt
[
-I
-2
."
i'''lr
P.,
fI]
-1
n
Dept. of EEE,SJBIT
Page 131 Dept of EEE,SJBIT
Page 132
Modern Control Theory 10EE55 Modern Control Theory 10EE55
If f;x) is negative definite. we see that \',xl is negative definite. Hence V(x)is a
8.7 Stability Analysis of Nonlinear Systems Liapunov function. Therefore, the origin is asymptotically stable. If FC \" - /'(x)f(x)
In a linear free dynamic system if the equilibrium state is locally asymptotically s.table,then it is tends ,to infinity as II.,II - ,'r .' then the equilibrium state is asymptotically stable in the
asymptotically stable in the large. In a nonlinear free dynamic system, however, an equilibrium large, .
state can be locally asymptotically stable without being asymptotically stable in the large. Example 8.12: Using Krasovskii's theorem. examine the stability of the equilibrium stare x
Hence implications of asymptotic stability of equilibrium states of linear systems and those of ::11,0or.the system gt ven by ~:i::t - -x, jJ ad ;\:::e- X'I - ::r...- :.~
nonlinear systems are quite different. .
If we are to examine asymptotic stability of equilibrium states of non linear systems, stability .
f(:r) -
[ r _
-):
y '__ .]
analysis of linearised models of nonlinear systems is completely inadequate. We must investigate t "'\">"I ".!,)
nonlinear systems without linearization. Several methods based on the: second method of
Liapunov are available for this purpose. They include Krasovskii's method for "<:sting sufficient The Jacobian matrix is given by
conditions for asymptotic stability of nonlinear systems, Schultz - Gibson's variable gradient
method for generating Liapunov functions etc. .
J(rasoyskii's Method:
Consider the system defined by ,;,- trr), where x is an n-dimensional vector. Assume
that f(0) = 0 and that f{x) is differentiable with respect to Xi where, I = i,2,3, ..... ,n. The
Jacobian matrix F(x) for the system is
I'(,K,} - F'Ct) + F(: ....).
ail ail cfl
aXl
ah cf:
ax: i);<.
afl
- [-'n - I-
'1:h'; + 1_' I -1 -
'n 1
1\';
oin eir. cf. This is a negative definite matrix and hence the equilibrium state is asymptotically stable.
X1 ax.. ch... fT:,,)f(xl - x; I (.~, x-, ~!)' "A ashll ,eA. Thereforetheequilibriu:,state is
Define = r '(x) of- /'(x). wl:el'eFF.(::tl) is the conjugate transpose .of F(x). If the asymptotically stable in the large.
Hermitian matrix f.(x~ is negative definite, then the equilibrium state x ~ 0 is
asymptotically stable. A Liapunov function for this system is ~'(x:- . f(x)f(xJ. If Example 8.13: Using Krasovskii's theorem, examine the stability of the equilibrium state
in addition f'(x)f(':e)"" ". IIxll then the equilibrium state is
x = 0 of the system given by
asymptotically stable in the large.
Proof: r:t".,..)_ [ -3,.,. + x,,'il ]
If !tx) is negative definite for all x 0, the determinant of f is nonzero everywhere )'I-.t2-X,~
except at x = O. There is no other equilibrium state than x = 0 in the entire state space.
Since f(0) = 0, f{x) 0 for x 0, and Vex) =f'(x)f(x1 is positive definite. Note that F(,,) = r -'l
l' _I_::I;:~
I I
I(x) - F(x)i< - p( r)f(r) .
We can obtain Ii' as f'(~) ..... Ff(x) 1 r(:~)
, "
Vex) - f'(x)f(x) .'
+ f'(x)!(x; '"' [F(x)f(_")!'f(x) + {'(x)F(x),fLy)
= [,(x)[F'(x) + F(xjlf(x;
'"'f'(x)f(x)f(x:
This is a negative definite matrix and hence the cq?ilibrium state is asymp.toticallystable.
'. 2) Determine state model for given transfer function (Jan 2007)
C(S) 24
". R(S) s3 + 982 + 26s +24
C(S) s2 +7S +2
R(S) s3 + 952 + 268 +24
C(S) 24
R(S) s3 + 982 + 26s +24
24
(8 +2) (8+3) (S+4)
UNIT 3 &4
1) What is STM? Obtain the state transition matrix using power series method (Dec 09)
(Jan 2010)
a) A =[ 0
-I
J
.I
-2 . b)A=[~ :J
2) What is STM? Compute the state transform matrix eAT. (Dec 2008) (June 2007)
. = [1 IJ
a). A I' b).A=[O -I IJ. c). 0
A=[ -2 I]...
-2 -3
Dept. of EEE,SjBIT
Page 135
Dept of EEE. SjBIT
Page 1
Modern Control Theory 10EE55 Modern Control Theory 10EE55
4 ) What is STM? Compute the state transform matrix eAT using Cayley Hamilton Example: Determine the state controllability of the system by Kalman's approach,
theorem. (Dec 2008) (June 2007)
0 1 1 [OJ u(t)
a A= 0 .,. I] [ ao ",Q-2
i;(t) = (J1 x(t) +
a~. A=[~ c).
. [ -2-3
i.
....3
. 1
A '" [~
o
~
3
~Il
by Cayley-Hamilton method.
Test the observability using Kalman's approach.
UNITS UNIT 6
1) Explain the concept of Controllability and observability, with the condition for complete 1) Design a controller K for the state model ( Dec 2009)
controllability and observability in the S- plane (Dec 2010) (June 2010) .
Example: Design a controller K for the stare model
2) Check the controllability of the system (Jan 2008)
X(I)= [~ ~ ]!5(ll+[rr(t)
Example: Check the controllability of the system. If it is uncontrollable, Identify the ShIle, Such Ihallhe system shallhnw damping ratio as 0.707 & senling lime as I sec.
which is uncontrollable.
. [-I -I] ~ [I]
X(t)= 0 _2.~t)+ 011(1)
2) Design controller K which places the closed loop poles at -4
Acermanns formula. (Dec 2007)
j4 (or a system using
Ex. Design controller k which places the closed loop poles at - 4 j4 for system
3) Check the controllability of the system by Kalman's method (Dec 2009)
X(A)=[~ ~]X(t)1o[~}(t)using Acennann's formula.
Exomple: Using Kalman's Approach delennine the state controllability of the system
described by 3) Design a full order state observer. Assume the eigen values of the observer matrix at
H3H2Y;Qitll (I) -2 j 3.464 and -5 (June 2010) (Jan 2010)
with XI :; y,x% ;;; )'-11
," - 3""2 -2""'I -"11
, 2 -- ..
>:(1)=
0
0 0
1
011 X(t)+'OCD'IU(t) 1) What is phase- plane plot? Describe delta method of drawing phase- plane trajectories
Y [1 0 o]X(t}
2) What are singular points? Explain the different singular points with respect to stability
Dc,ign" ,("I<" r""d back controller which place Ih,' do,,,,.1 "><'1' pole'; at 2 j3A64und
5, Give the bk"l\:k diagram nf the contro] configuration of non-linear system (Jan 2010) (Dec 2009) (June 2009) (June 2010) (Jan 2009)( Dec 2010)
4) Design a full order state observer. Assume the eigcn values of the observer matrix at -
3) Explain the common physical Non-linearities (Jan 2010) (Dec 2010) (Jan 2008) (June
2 j 3.464 and -5 (June 2010) (.Jan 2010)
2010)
Example : Consider the state model
5) Design controller to take place closed loop poles -I jl, -5. Also design an observer such
that observer poles are at -6, -6, -6. (Jun 2009) (Jan 2007)
"
",
MOdern Control Theory 10EES5 MecleFR CeRtFel Theo:Fy
~ s2 +7S +2
R(S) s3 + 982 + 26s +24
2) Determine state model for given transfer function (Jan 2007)
C(S) 24
R(S)
.'
.'
.'
.usls '333)0 '~d"a
(tH) (f+s)(Z+ s)
tZ
t'l+ s9(: + zS6 + ~ (S)H
tZ (s)~
o1+~~J
a t~J Lo~x
~;]
r~-:-: ] 0~ 1 ~- UI.I0J.I01~llJUI oq 01 SI :11JO .I01m,!LUOllJp~q1
(600, aunr)mJ0.lllUlPIl3SIlJ ul lapom J)IlIS B dopAaa (."
0 tJ'~ xXJ
J!. ;:] co;)
~~t Atz+rXZ - =r~
~ X.[1
J x r _rX =ZX IXi:+ZXL+''X=
IX Z + IX L + IX = (1)J
1x~-ZX=1~
JX Z + IX sL+ IXZS= (S):)
(S)TX
(S);)
1 X 6-9(:- tz- .;
(l)a 0 + zXOJ 1 0 0 J z~
[tX]
t+s o Ix 0 1 0 l~
=-1- lJ ~ l r~ = l.~
r-
L_ ("H) (+8) (Z+s) rX ='lX= l:x
(S)lX J (S)'X T (S)fX --
T
~ tZ
(S)H 'lX= l~
~
IX
SS3:3:01 A.IOcH{l. [O.IlUOJ U.IapOW
SS3:3:01 AJOal{l [O.I:lUO:>urapojq
Modern Control I heory lOEl!55 MOd'eI Il'6oIltl 01 'fheoI Y IDEESS
UNIT 4
1) What is STM? Obtain the state transition matrix using power series method (Dec 09) (Jan 2010)
=[~ 0]+[1 1] [I
2 ,
t:
a)A=[O -1 -21], b)A=[1 0 1I] 1t+ 0
1]1 21 + .....
1
A2t2 A3f2
= I +At+--+--+
a) J(t) .
+-+ ...
3
2! 3! r r
l+t+-+-+
'l .
...
, t3
l r+ l "
=[~ ~]+[~l_12}+[~1 2! 3!
t2
2
t3
t3
1--+-+...
t2 t-t 2 +-+... 1
t3
1+1+-+-+ ...
2! 3!
_ 2 3 2
- 2 (3 3t2 2t3 el
l -t+t --+ ...1-2t+---+
2 2 3
... .r:::
.- [ p
= [e-' +. i:
- te -I
b).
,
A=[O .1 ]=>[SI-AJ=IS
- 1 - 2
-1
-t41:-----+2(Sf--=-t-F-02&)J4-J
1 _
..
.
[S+3
-2
1]
S
S+2
=> J(t) = ,t-1 J(t) = :-1 (S + 2)(S + 1)
s -1 ]-1 = (S +!__.)2 (S 1
+ -]1)
=> (t) = LSI - Arl = [I S+2 -I . 2e-I -e -21 e-I -e -21]
[
(S + 1)2 (S: 1)2 = [ - 2e-t + 2e-2t - ' + 2e-21
(1+ t)e-I
(t) =-1 (t) = [ _ ie:
(Jan 2006)
=> f(AI) = ~IO = ao +al~ 2). Find S:rM by Cayley Hamlton method, given:
i0
=> (-1) 10 = ao - at 1I) A =[1
0 :] b) A=[O
-I
_I..,
-.J
1 CI A=
L- 2
_13]
Hence, ao = I + at = 1-10 = -9
C(S) = 82 +7S +2
4(8) Xl s-1 s-1
~=
= '8+"-:- =
1 + 4 s-1 1-(-4s-1)
X2
.. .
C(t) = Xl + 7 Xl + 2 Xl
24 -1 1 s-1 1 s-l, 1
R(S~ ~X3
-2
'0'0
X X2~Xl '
X3 ..
-3
' 2
-'I,
, OC(S)
Xl =X2-4 x t
X2= X3- 3 X 1
3. Cascadingform
X3= -2X3+24V
tfJ:!edenominator ofTF is to be in factor.fonn
-_
C(S) _. __
R(S)
2_4
s3 + 982+ 265 +24
_
c= [1 0
24
_ (s +2)(.5+3)(st4)
1
(s +2)
X (8) -L., X2 (8)
(S+3)
[xj [~4 1 OJ
X2
Xl
0 -3
0
1
() -2
[~l ~Ol
X + Q R
X3 24
.15 ..
16
0 0...... 0 1
Such that the syst~~ 'shall have d~~ping ratio,as 0.107 & setlling time as 1 sec:
Solution: .. ," ...... ,.,': ., " ,. ,
.'. 'a)' Ct:mvefting thed~ireri specifications as locations'of closed loop poles at - 4 j4
=[_:. -L '-~ _:. (6)
(1)
,',
. . ','K"[K'
= l' 'K2l-' . '.~~:_'.,-
.. J;' -:-::--;:-:',7"" ',' . ":" From Equ.(3),
oX From given state model " . x, = P l1x,+P'2X2+ +P'nxn= P,x
Taking the: derivative
.,,"[A;-:~~lJ[O.1]:_[ 0'" J
:OJ. .r X, = P,x(t) = P,[Ax(t)+ BU(t)]
, ,\'.:,. : ,.ll0, 0. ...K1;: K2 . ",. ,"
= P,Ax(l)+ P,Bu(t)
From Equ(4)
(8)
.- x=x
2+0 (9)
Comparing (8) and (9), P,B= 0
& x 2 =P,Ax(t) (10)
Taking again derivative of Equ(1 0)
18
VTUJe.
.
'. VTU~
Learning Learnfng
x = P,A[AX(t)+ Bu(t)] ----------------------------------------------
Hence the corresponding characteristic poly will be
=P,A2X(t)+P,AB andfromEqu(4) ISI-(A-BKI=sn +(0, +Kn)S'-' + ....... +0. +K, (14)
and characteristic equation will be
x=x.+o=> x. =P,A2X(t) & P,AB=O
Sn+(a, + Kn)Sn-, +(02 +Kn)S .......... +on +K, = 0 (,1_5)
Similarly xn = P,An-'X(t) & P,An-2B =0
Where K"K" ..... K. can be chosen real numbers arbitrarily.
Taking again derivative
Let the desired characteristic equation be
P,An-'B=1
l
sn +a,sn-' + ....... +an =0 (16)
Thus,
Equating the coefficients of like powers of S from equation (15) and (16)
1[ 1 a, = a, + Kn => Kn =a,-o,
- r P,X
P,AX
X = P,A-;'_'X =
With P,B=P,AB=P,A2B=
P,
P,A
p,~n-,
X
.... P,An-2B=0
, (11) a, = a, + Kn_1=> Kn_,=a,-o.
t:-~
3) Compute A & B as
0 ....
:_fltl
A=PAP-t
0 1......
(13) B=PB
4) Assume the controller K
-0n-,-K. -o,-Kn
5
4
.. 19
20
VTU_2
.
Learnrng
K = lK,K2....KnJ and
K= [0001)0, -'<p(A)-------c-----(4)
1 0
010 Where
Determine tA_SKJ=[ ~_
cp(A)= A' +a,An-' +a2An-2 +a3An-3 +....... +8nl (5)
-an-K, & Q, =[SI ABIA2BI.......An-'B] (6)
5) Determine the Cba equation as
Design procedure:
ISI-(A-SKI=sn +(a. +K,)Sn-' + +a, -K. =0 I. For given model. determine Q, and test the controllability. If controllable.
6) Determine the desired cha equation as 2. Compute the desired characteristics equation as
S" +a,S"-' +a Sn-2+ +a. =: 0 sn +a,S"-' + .........+8n =0
and determine coefficients a" a,.: a,
7) Determine the elements of gain matrix K.K,.K2 K.
3. Compute the characteristics poly of system matrix A as
as K, =a.-an <Il(A)= A"+8,An-' +a2An-2 + .......+8.
K2 = an-,-an-, 4. Determine Q;' and then controller k will be K=[O0 .... 0 I] Q;'. <1>( A)
K. = a, - a, and hence This completes the design procedure.
K=(K,K2 KnJ Ex. Design controller k which places the closed loop poles at - 4 j.4 for system
8) TIle feedback controller K for original system is
K=KP X(A) = [~ ~ ]X(I)+[~]u(I) using Acermann' s formula.
This completes the design procedure
Answer: Check for controllability
Design of controller by Ackermarm's formula:
We have
0, = [BI ABJ=[~ ~] Its controllable as f(Q,)=2.
6 .' . 7
" "21 22
VTUJe
.
Learntng
X(I)= Ax(l)+ Bu(t) 2. Desired characteristics poly is
Y(I)= CX(t)with (S+2+j3.464)(S+-j3.464)(S+5)
u(l) = -KX(t)where =:;S3+9S2 +20S+60
K =[K,K2 Knl
Comparing with S3 +a.s" + a,s + a3 we obtain
From above relations, the closed loop system is Q, =9, a2 =20, Q3 =60
P=[::A
" P,A""
l; P, = [0..0...01 1 O~'
X(t) =[ ~ ~ ~
-6 -11"-6
lX(t)+[~lU(t)
1"
Design of state observer:
Design of observer: The device that estimates we state variables with measurement of
output along with input u(t) is called state observer. -
Y = [1 0 O])(t) H all the state variables are estimated, it is called Full-order state observer. If only few
statevariable are estimated (which can't be measured accurately), the observer is called
Design a state feed back controller which place the closed loop poles at - 2 j3.464 and Reduced-order state observer.
-5. Give the block diagram of the control configuration
Design of Full-order observer: Inthis case all the state variables are to be estimated.
Solution: By observing the structure of A and B matrices, the model is in controllable
conmical form, Hence controller K can be designed. " a) Open loop observer: Here it is assumed that all the srate variablesare estimated
1.
"
8 -1
ISI-AI= 0 8 ~1
16 11 8+6
I~ 8' +682 +118+17
"
correctly and initial state are also estimated correctly.
i.e For State model,
X(t) = Ax(t)+Bu(t)
Y(t) = ex(t);
initial state X(O) are known and
Comparing with S +0,82 +02S+0., we obtain 0, = 6,02 = 11,0. =17 X (I) is estimated states and
--5(0) IS estimated lrutial states !he resulting observer in open loop observer.
8
9
24
A
If 'm' is chosen properly, (A-mc) have reasonably fast roots, then X (I) will decay to
';((t) zero irrespective of X (0) i.e X (t) = X (I), regardless of X (0). Funher it can be seen that
equation (5) is independent of control input u(t). *"
Draw backs of open loop observer. The obserVer gain matrix 'm' can be obtained in exactly the same fashion as controller
I. Precise Lack of information ofx(O) gain matrix 'K'
2. If X(O)i:X(O), the estimated state X(t) obtain from open loop estimator will have De!ligT)procedure:
continuously growing error and hence the output, I . Let the desired pole locations of observer error roots be fJ" fJ" ......fJ" then the
3. Small errors in A,B, and disturbances that enters the system but not in estimator 'corresponding poly equation will be
will cause slow estimate process. .
(S-I3')(S-132) ....(S-I3,)=0 (1)
Hence better to go for closed loon estimator.
~ S' +a,S~-1 +a2Sn-2 + ..... a, = 0 (2)
Design of closed loop Full-order state observer: Figure shows closed loop FuH orcter-
determine co-efflents ai' i12 , .. all
state observer (Luen berger state observer) .
Z... Let the pbserver gain matrix be m
m'=-,im"IJl., ....... mnlT
obtain the characteristics poly equation of the closed loop system with observer as
I~HA-mc)I=O
S' +(0, +m,)Sn-' +(0. +mn_,)Sn-2 +.......+(0, -rn.) =0
=:;. (3)
3. Comparing the co-efficient of like powers of S from equations (2) and' (3)
I a, =01 +rn, ::::)mn =a1-u1
I
I an =On -rn, ==> rn, =an -an
L_____
I m I
I where the co-efficients a" a, , a. are obtained from poly equation
"-=- S~A~E-
:B~E:V:;- - - - - .J
ISI-AI=O
=:;. S' +o,sn-l +.......+0,=0.
4. The observer gain matrix 'm' will be then
State model is X(I)= Ax(l) + au(l) (1) m=[m"m2, ....... m,lT
Y(I)= CX(I) (2) Example: Consider the state model
Estimator output is V(I)=CX(I) (2a)
11
.'
..25 26
~I)'[~0~~]X(I)+m
Required when few of state variables cannot be measured accurately.
Design of Reduced-order state observer:
In this design procedure. let us assume one measurement Yet). Which measures one state
X,(t). The output equation is given by
Y(t)=[O 0 1]X(I) Y(t)=CX(t) (l
1) Desired characteristic equation will be Where C=[1 0---0)
Partitioning the Slates X(t) into Slates directly measured X,(t) and states directly non
(S+2+j 3 .64) (S+2-j3.64) (S+5)=0.
measurable X,(t). which are to be estimated.
=> 8'+98'+208+60=0
hence comparing with S' + a,s' + a,8+ a3 = 0 . _f
Xl(I)]
r.e X(t)-~:e.e(~I)~ _
('I
a, =9, a2 =20, a, =60
Partionionly the matrixes accordly & state equation can be written as
2) Let the observer be m = [m" m2 ........ m. l'
X~(I)]=(al1 a,.][X,(I)]+[b']U(I)
0 0 -6] [0 0 m,l [ X.(I)
[A-me]= 1 0 -11 - 0 0 m2J a., aee X.(t) b.
[
o -6 0 0 m,
and Y(t) =[1 0G:\::] (4
[A-mel=[~
o
~ :(;~::'!)l
-(6+m3)
from equation (3)
X. (I) = a .. X.(I) + a.,X(I)+ boo (5
KnownlnpUI
~181-(A-mc~=0
from (4) Y(I) = X,(I)
~ S3 +(6+m3)8' +(11+m.)S+(6+m,) =0 => Xl (t) = Y(t) Also from
Comparing with
Bquation (3) X, (I)= a" X,(I) + a,.X. (I)+ b,U(I).
Collecting the known terms.
S3 -I'O,S2 + o.s + 0. =0
0, =(6+m3).0. =(11+m.),03 =(S+m,)
Y(I)- a" X, (I)- b,u(l) = a., X. (I) (6)
(II) Knownlerms
3) From I and II. Comparing equation (5) & (6) with
9=6+m3:::>m,=3 )(1)= AX(I)+bu(l)
20 =11+m. :::>m, = 9 =
and Y(I) eXIt)
SO=6+m, :::>tn, =54 we can have similarity as
X(t)~X.(I)
4) Hence m = [54 9 31T
A(t)~a ..
bu ~ a.,X,(t)+ b.u(t) (7)
c=a,.
Y = Y(I)-a"X,(t)-b,u(l)
Using Equation (7), the state model for reduced order state observer can be Obtainedas
~.(t) = a )<.(I)+ a.,Y +b.u(I)X,(I)+ m(Y(I)+ (Y -allY -b,u -"-,.:>:.(1) (8)
12 13
:
... 27. . ..~ 28
I
I (same as X(I)::A)I)+ bu(l)+ m(y(l)- y(l)) where y(l)::a,.X.(I)
! ~~~ b d
Defining X. (I):: X. - Xe : From output Equation it follows that states X,(t) & X2(~)cannot e measure , ~s
corresponding elements of Crnatrix are zero, Let us write the state equation as
~ ~.(I):: X.(I)-~.(I)
I Substituting for X. (I) & ~. (I) in above equation & on simplification, X'j :: [01 00 -24
)(, J
0 l[X,(t)]
X,(t) + [50]
2 u(l)
I X(I)::(a.. -ma,.)X. [ X 3
0 1 -10 X3(1) 1
Corresponding characteristic equation is
ISI-(a .. -ma,.~::o (J a.. =[~ ~J.ae, =[-~4J. a'e =[0 1)
Comparing with desired characteristic equation 'm' can be evaluated same as in full-
order observer.
I To implement the reduced-order observer,
Equation (8) can be written as
au = [-.1o.1~e = [520]b, :: [1J
~m=[1;~] ~
Hence state model of observer will be
(2)
lo 1 -10J 1 Assume the full-order observer model be
Yel)::[0 0 1JX(I) k(t)=AX(t)+Bu+M(y-CX(I . (3)
To design the reduced-order observer for un measurable states such that the given ~alues Hence the state feedback control law using the estimated states WIll be
will be at-IO. -10. . U(I):: --KX(t) (4)
14
15
..30
VTU)2:
Learn~n
--0-_
Using equation (4) in (I a) Taking LT.
'1(t)
Solution:,System is observable & Controllable. Using the procedure of controller design
& observer design, the controller will be
K=[4 I I]and observer will be
M=[12.~. -71IT
..
r Transfer fynction of uncompensated system will be
Mu(S)=qSI-Ar'B.
, 1
1 MUIS1 S3+6S2+11S+6
I Co"t"olllf"T ObSe:rve'Y l-(_ Transfer function of compensated system will be
L 1 Compew s,,for
Me(s)= U(S)=K[SI-(A-BK-mCJr'm
Y(S)
Transfer function of compensator: M S~ S~+11'gS+618
e( )- S3+19S2+121S+257
Now X(I)=AX(I)+Bu(I)+m(y-cx(I)) (1)
U(I)=-KX(I) (2)
From (1) & (2)
16 17
.3L.
32
.. , ....":(2.-' .'
" ,':
VT~
Le~rnfn~
TakingL.T.
Solution of State Equations
SX(s)= (A-BK-mC)X(S)+mY(S)
After obtaining the various mathematical models such as physical, phase anc
~ mY(s)= (SI-(A -BK -mC)jX(S) (4) canonical forms of state models, the next step to analysis is to obtain the solution of stat,
Taking L.T. of equation (2) ------equation.Flom die solution 01 tlie stale equation, -the transient response can then be
U(S)= ;"K.X(S) obtained for specific input. This completes the analysis of control system in state-space.
(5)
From equations (4) & (5) The solution of state equation consists of two parts; homogenous solution and forcer
solution, The model with zero input is referred as homogenous system and with non-zen
~ U(S) =K(SI-(A-BK-mCr'jm input is referred as forced system. The solution of state equation with zero input is callec
-ViS)
as zero input response (ZIR). The solution of state model with zero state (zero initia
=D(S)
condition) is referred. as zero state response (ZSR). Hence total response is sum of zm
andZSR.
.
:
"",
rt
..
(01_ 'I); =
.......+~~("z+'ty+[=("z-'ir/A..~-'t~
,!.."I+ 'I) t'z+ 't)~
,a oJ iz
[ ,a; ,a = .... + +(OI-'I)V+[=(OI-'I)'"
(1_ 'I),Y 'I'
... i z iZ
+--;-+-+1+[ ....+ +('I-'I)Y+I=('I-'I)'" 'WUd
(" '(.1 ('I-'I),V 'I' ,#'
Z ... i iZ ('I)r('I- 'I); = ('I)r
[ "'+,+,1+1 +-+-+1+[ 1=
r ,I z (0')r(01- 'I); = ('I)r
"~'! ,ll, 01 ,II, wOlJ pUll ,II, 01 ,0), wOlJ UOll!sU1!Jl01 [Bnb:lS!! (OI)r(O,- 'I); = ('I)r
..... +~ [I Otl[1 0J+[I 0J= '. "101 ollIIoll uO!l!sU1!Jl:ll{l
.;)., '(SUQ!l!SU1!Jl
[B!lU:lnb:ls
,1, 1 IJ 1 1 <l I
J? J:lqwnu.1lOIU!p:lp!AlPoq UIlJsosscoord U0!l!SUlIJI ;)IBISIBl{lS:!Hdw!A)J:ldoJdS!IlL)
i'l
.
......+_+_1_. -+IJI+/=(I);
(q : (01_ 'M = (01_ '1);('1_ 'I); '(17
,I,Y ,I,Y'
('1+ '1);=
,_dl- . ,_dl- .......
+~~('t+~y+[=('t)/!(~~
[ l_a1 ,_al+ ,_a ~) {1+ 1)~
J=
tz+
i iZ
....+-y;-;;+--y;-;;+ 'W+ 1=('I); pue
"'+__-_+17 ,,- 1+1- . ,I,V, I,V
z:
(....r z 1
Z
r ...+ --Z1
'",
1 iE iZ .
[ ".+_+ I_I .,,+ + Z .... +-,-+-,-+ 'W+I=('I); :joOJd
,I z J -;I-I ,',Y ,I,V
:('1+ 'lltP= ('I)('Ij .(
.... +~~ [z- I-]+~ [z- I-]+I[Z- I-J/I 0J- (I) ,_; = (1-);
, ,I 0 ,I, 1 0 1 lo I -
,_; l{l!A\S:lp!Sqioq jjUJAld!I(DUI-:IJd
1=(1-);(1);
'(H; pUB(I); jjU!Ald!llnw
....+~-~+W-I=(I-);<=
,I,V ,I,V
.... + --iZ + IV + 1 = (I); J~P!suo::>.joOJd
,I,V
: (l-ltP = (I),_; ,(Z
'[=(0)$ OIU!sllns:!J 0=1jjU!llll!lsqnS
(1);=
i iZ
....+--+--+IV+[= <va
,I,V ,I,V
'U0!l!olPP Ql{lwGij SMOI\OJ JOOJd/=(0)$ .(~
:(1):;:, Xl.lllIW U0!l!SUII.IJ. 31lllS JO S3!lndo.ld
6UIUJIt)'cq
\!) -:
~n~A
-. .
1 .~
V
35
36
VTU~
LecirnTng
Example: For system matrix A =.[ 0 2 ] show that the STM is
-2 -5 b). A= [ 0 1] =>[SI-A)=
-1 -2
[S
1 (S+2)
-1 J
4...
3
1 ...,
-e --e
3
-e
2 _, --e
3
2
3
""'1 S+2
-2 _,
[ -e +-e
3
2
3
...
,
1 ...4
-e
3 3
-e -" J => (I)'= [SI- Ar' = [~ -1
S+2
J-' =
[
(S+I)'
-1
(S+I)
I 1
I (S+I)' (S:I)2
2).Inverse Laplace Transform Method:
0 + I)e-' le-' ]
Consider the state equation with zero input as (I)=I(t)=
X(I)= Ax(I); [ -ie:
(I-t)e-'
X(IO)=x(O)
xes)=[SI -Ar'x(O)
Taking Laplace Inverse on both sides.
X{I) = :.e-'[SI-Ar'x(O)
.'
Hence.(I)
[S_+/
= (S + 2)(S + I)
~J
Comparing the above example with
X(I) = eA' x{O)
It shows that
eM = STM = (i) =:.e-'[SI - Ar'
. .
=> (I) =.I(I) =;..1(S+2)(S+I)
[S_+23 ~J
~ (I) = ..,[SI -Ar'
Example:
Obtain the STM by Laplace Transform (Inverse Laplace Transform) method
vru~
.
LearnIng
=[~ ~,']
b) A = [ 0
-1 -2
1 J; characteristic equation is IAI _ AI = 0 ~ /A1 -1 '
A+2-/
-0
~4,=-I,A,=-l.
For second order system, Deterirnine the S1M of system matrix
eh =ao+a".t (I) 2 1 4]
A = 0 2 0 by Cayley-Hamilton method.
.
~ e""=ao +a,4, [o 3 1
::)e-'=ao-a,; (4,=-1) (2) Now characteristic equation is.
Differentiating Eq. (1) with respect to A and substituting A=A2=-I
al =Ie-l IAI-AI =IA~2 A-_I2. -04 =0 1
8
9
42
VTU
-----------------,------_._-----
LearnTnc
Solution of Non-homogeneous State Eguation X(I) '" [
-I 11
JX(I) + JU(I)
[Ol
-I -10 10
Consider the state equation with forced input u(t) as
X(I) '" Ax(I) + BU(I) ;
y(l) ",[I O]x(I);X(O) ",0
(I)
X(I.)"'X.
Solution: The state equation solution is
~ X(I) - Ax(I) '" BU(I) X(I) "'eA'x(O) + J eAe'-<)Bu('t")dr (I)
Pre-multiplying both sides by e-" ,
, Since x(O)=0,
e-A' [X(I) - Ax(I)] '" e-" U(I) (2)
Consider, X(I) = J eAeH) Bu(r)dr
~[e-" X(I)]
dl
~ _Ae-A,X(I) + e-A' X(I) = f (I-r)Bu('t")d't" (2)
~ e-"[x9TO-!i.x(I)] (3) Now, the state transition matrix (I) can be evaluated as
hence Eq (2) can be written as eA' =(I)=.:c,[SI-Ar'
~[e-"x(l)l'= ,,"'U(I)
dt =[1.0128e-." -0.0128e-" 0.114e-" -0.114e-" J
Integrating both sides with time limits 0 and t -O.l14e- Q" +0.114e-" -0.0128e-a" +1.012e-o,o
e-'" X(I) I:", j e-A<Bu(r)dr 1.0128e "-') -0.0128e-"(o-') 0.114e-o,(o-,)-0.114e-'(o-') ]
~(I-r)= [ -0.1I4e <O-<) +0.1I4e-,e,-,) - 0.0128e-O,,,-<)+ 1.912e-o,e,-,)
Hence
. I-'
$ .
s -I
.
=.:c 0'
[2
"
S+3.
I 10 11
I
iL__~.
44
Hi _, -e -2, --e
-e 4 _,,]
_ 3 3
- [ -5 e-I +e-21 +~e-41
3 3
and yet) =5e-' - e-"
1 '""
="2\,;-e _?, _".
tr-e."
= !(l- e-2'l)
2
13
'.
.. 45."". 46
VTUg
Liarn7n
Controllability and ObGervabillty Independent 'if C matrix, tjte controllability of the system is frequently referred to as the
Concept: controllability of the pair [A, B).
Consider the typical state diagram of a system. Example:' Check the controllability of the system. If it is uncontrollable, identify tile state
which is uncontrollable.
X(I)=[~I ~~]X(l/+[~}(t)
Solution: First, let us convert into diagonal form.
. 1).1 ~ _/A.;I) , 1 1_
-nt- O. A~~-O
The system has two state variables. XI (t) and Xi(t). The control input u(t).effeclS =>(A.+ I) [A.+ 2)=0
the state variable XI(t) while it cannot effect the effect the state variable X2(t). Hence the
state variable X2(t) cannot be controlled by the input u(t). Hence the system is
=>A' +3A+2=O
uncontrollable, i.e., for n" order, which has 'n' state variables, if anyone state variable is => (A. + 2)(A. + 1)= 0
uncontrolled by the input U(I), the system is said to be UNCONTROLLABLE byinput => A.=-2,A =-1
u(l). "
Eigen vector associated with ..I. =-2
Definition:
For the linear system given by -2. II => _11J
-L
VI
X(I) = AX(I)+ BU(I) 1 001
Y(I) = ex (I) + Du(I) Eigen vector associated with A.= -I
is said to be completely state controllable. If there exists an unconstrained input vector -1+1' 1 1=>10 11
u(t), which transfers the initial Slate of the system x(to) to its final state x(tr) in finite time
f(tr-to) i.e. fr. It can be seen if all the initial states are controllable the system is completelj
1 o -1+2 0 1
as
~[Q'-2J[1 IJ=[-20-1 0]
j
X(I) =eA1 X (0) + ,,'(1-<) BU(1')dT. Assuming without' loss of generality thatX(t)=O"the
I'~IIO
solution will be
o ~
. P-IB_['O 'JIJ_'[ 111
-I I 0 -IJ
- X (0) = j e-M Bu(t')<i1'
As,or1 B vector contains zero element, the system is uncontrolled and state XIii) is
o uncontrolled.
The system is Slate controllable, iff the none of the elements of B is zero. If any
element is zero, the corresponding initial state variable cannot be controlled by th~ input
u(t). Hence the system is uncontrollable. The solution involves A, B matrices and is
14
.41 _ 48
VTU)e
LecifnTns
2) Kalman's approach:
This method determines the controllability of the system. Verify the result with Gilbert's approach.
Statement: The system described by Solution: Kalman's approach: Here,
x(t) = Ax(I) + BU(I)
Y(I) = Cx(t)
is said to completely state controllable iff the rank of controllability matrix Qc is equal to
order of system matrix A where the controllability matrix Qc is given by, .
BIA2BlAB
Q, = [B lABIA2BI..... IA"-IB].
o
i.e., if system matrix A is of order nxn then for state controllability
p(Q,)=n
~Q<=[: -3
1
~
i, =X2 +u
[0 1Tx,] [1] 1 [~ 1
X, ]
[ x2 .:: -2 -3..._x, + -2 U(I)
l
2(1)= 0 -Io,
o 0 -2
0 Z(I) + -1[."(I)
_l_
r
:=)A= [ 0 1 ] ,B= [ 1 ]. ,_:. 'I 2
-2 -3 -2
=,[~
Now AB=[ -20 1
-3 -2
I]=[-2]
1
4
Here, B
Observability:
-1 ~ has no zero element in any row, hence system is controllable. ....
Concept:
~Q, =[BIAB]=[ 1 -42] A system is completely observable, if every state variable of the system effects
-2 some of the outputs. In other words, it is often desirable to obtain information on state
variables from-the measurements of outputs and inputs. If anyone of the states cannot be
IQ,I= 1-21 -21
4 =0 observed from'the measurements of the outpits and inputs, that Slateis unobservable and
system is not completely observable or simply unobservable.
hence rank p(Q,) is <2 and system matrix A is of 2x2. Therefore system is not Consider the state diagram of typical system with state variables as x, and X2 and
controllable. y and u(t) as output and inputs respectively,
Example: Determine the state controllability of the system by Kalman's approach.
X(I) =[~~
o
~lX(I)+[~}(I)
-2 -3 1
u~~.~(')
""--~
J
-1
16
17
50
VTU~
Ledrn1n!
On measurement of y(t), we can observe XI(t)but we cannot observe X2(t)by
measuring out y(t), hence state is not completely observable. . 1.1=0~p(!'.'<2
Definition: 'j; 2. ~o, system is unobservable.
Order of system
A system described by Example:Show.that the lystem
itt) = A.%(t)+ Bit(t) .
yet) = Cx(I) + Du(t) is said to be completely observable iff with measurement of output
yet) over finite time interval t.::;t.$t(along with input u(t); ~Io the initial state x(lo) can be
estimated. .
i(l) '"[ ~ ~
-6 -II
~
-6
lX(t) + [~Jl(,)
I
Explanation: yet) =[4 5 t]x(t) is not completely observable by
Consider the output equation I) Kalman's approach
yet) = CX(I) + Du(t)
2) Gilbert's approach
Assuming D=O. y(t)=Cx(t).
Solution: Kalman's approach
All the state variables x(t) can be estimated with measurement of yet) provided C
matrix do not contain the 'zero' element. If any element of C matrix is zero, the .
corresponding state cannot be estimated and hence system is unobservable.
Kalman's Altproacb:
The system described by the model
A=
[
0 I 0] ~AT =[0
0
-6 -II
0 I
-6
I
0
~~I~l
-6
i(t) = Ax(t) + Bu(t)
yet) = Cx(t) + Du(t)
is said to becompletely observable iff the rank of observability matrix ,p. is equal to the
C=[4 5 I]~CT =[:j
order of system matrix A where,
. =[CT::?ATCT:AT'CT ... AT-'CT)
For n'" order system, rank of
i.e. p(.)=n
,p. must be n. ~A'c"[}"c'rl
It involves A and C matrices. The pair A, C is observable.
Example: Consider the system ~ . =rC:i--A;t--A~-~T1~I.I =0
i(t)=[:-02 ~J(t)+[~}(t) 5 -7 5
I -I -I
y(t)=!1 O]x(t)
Test the observability using Kalman's approach. ~ p( 9.) < 3 and order of system is 3. Therefore system is unobservable.
A=lo'-2 0];C=[I 0] Transferring the system model into observable canonical form with usual
procedure, the output equation will be
.. -1
=
yet) [0 - 2 - 2]z(t)
As C matrix contains the element '0' the corresponding state Zt cannot be
~ ATCT =[-02 ~II~]=[ -02] observed:Hence the system is unobservable.
.51. S2
VTU~,
LicirnTng VTU.Ae
'.
Lecirn7ng
CONTROLLERS
The controller is an element which accepts the error in some form and decides the proper Integral Controller:
corrective action. The output of the controller is then applied to the system or process. We have seen that proportional controller can not adapt 'Withthe changing, load
The accuracy of the entire system depends on how sensitive is the controller to th~ error conditions. To overcome this situation, integral mode or reset action controller is used.
detected and how it is manipulating such an error.. The controller has its own Jof:jc to In this controller, the controller output pet) is changed at a rate which is proportional to
handle the error. 'Ihe controllers are classified based on the response and mode of actuating error signal e(t). Mathematically the integral controller mode is expressed as
operation. On the basis of mode of operation, they are classified into Continuous and pet) = K, f e(t)dt + Po. The constant K, is called integral constant. The output from
Discontinuous controllers. The discontinuous mode controllers are further classified as the controller at any instant is the area under the actuating error curve up to that instant.
ON-OFF controllers and mulliposition controllers. Continuous mode controllers, If the error is zero, the controller output will not change. The integral controller is
depending on the input-output relationship, are classified into three basic types nanied as relatively slow controller. It changes its output at a rate which is dependent on the
Proportional controller, Integral controller and Derivative controller. In ID8DY practical integrating time constant, until the error signal is cancelled. Compared to the
cases, these controllers are used in combinations. The examples of such composite proportional controller, the integral control requires time to build up an appreciable
controllers are Proportional - Integral (PT)controllers, Proportional - Derivative (PD) output. However it continues to act till the error signal disappears. Hence, with the
controllers and Proportional- Integral- Derivative (PIO) controllers. integral controller the steady state error can be made to zero. The reciprocal of integral
The block diagram of a basic control system with controller is shown in Figure. Th'; error constant is known as integral time constant T;. i.e., T; = IlK;.
detector compares the feedback signal bet) with the reference input r(t) to generate an Derivative Controller:
=
error. e(t) ret)- bet).
In this mode, the output of the controller depends on the rate of change of error with ....
Conlroller oulpul respect to time. Hence itis also known as rate action mode or anticipatoryaction mode.
(p)
The mathematical equation for derivative controller is P(t) = Kl ":;') + p . Where K,
I{l) e(l)
Conlroller ..
I
I--;_-,-l Processto
is the derivative gain constant. The derivative gain constant indicates by how much
be controlled percentage the controller output must change for every percentage per second rate of
~)- change of the error. The advantage of the derivative control action is that it responds to
E_ the rate of change of error and can produce the significant correction before the
detector magnitude of the actuating error becomes too large. Derivative control thus anticipates
L/-,,-- -I Feedbackelemenl 1-------' the actuating error, initiates an early corrective action and reur's to increase stability of the
system b}' improving the transient response. When the error is zero or constant. the
Feedback
signal derivativeconttoEer output is zero. Hence it is never used alone.
Proportional Contretlerr
PI Controller:
.
This is a composite control mode obtained by combining the proportional mode and
In the proportional control mode, the output of the controller is proportional to th~ error integral mode. The mathematical expression for Pi controller is
e(t). The relation between the error and the controller output is determined by a constant
called proportional gain constant denoted as Kp. i.e. pet) Kpe(t). = . pet) - Kpe(t) I s, J e(t)dt 1.1'0
Though there exists linear relation between controller output and the error. for zen! error
the controller output should not be zero as this will lead to zero input to the system or The transfer function is given by 1(. I i
process. Hence, there exists some controller output Po for the zero error. Therefore
mathematically the proportional control mode is expressed as pet) Kp.e(t)+ p.. . =
The performance of proportional controller depends on the proper design of the gain Kp.
As the proportional gain Kp increases, the system gain will increase and hence the steady
.state error will decrease. But due to high gain, peak overshoot and settling time increases
and this may lead to instability of the system. So, compromise is made to keep steady
state error and overshoot. within acceptable limits. Hence, when the proportional
controller is used, error reduces but can not make it zero. The proportional controller is The advantage of this controller is that the one to one correspondence of proportional
suitable where manual reset of the operating point is possible and the load changes are controller and the elimination of steady state error due to integral controller. Basically
small. . integral controller is low-pass circuit.
The PI Controller has followin!-effects on the system.
15
16
,53 S4
Example: The figure shows PD controllerused for the system. Determine the value of Td
such that the system will be critically damped. Calculate the settling time.
18
"
ss
56
VTU)e
LicirnTn.s VTUX
LE~'cirnTn!
Chapter 6 response follows the .curve through the A, B and C, but at C an increment in frequenc
NON LINEAR SYSTEMS results in discontinuous jump down to the point D, after which with further increase i;
frequency, the response curve follows through DE. If the frequencyis now decreased, th
6.1 introduction response follows the curve EDF with a jump up to B from the point F and then th
response curve moves towards A. This phenomenon which is peculiar to nonlinea
Many practical systems'are sufficiently nonlinear so that the important features of their
I
systems is known as jump resonance. For a soft spring,jump phenomenonwill happen a
perfonnance may be completely overlooked- if they are analyzed and designed through shownin fig. 6.2(c).
linear techniques. The mathematical models of the nonlinear systems are represented;by
nonlinear differential equations. Hence, there are no general methods for the analysis lind
... i
synthesis of nonlinear control systems. The fact that superpcsitlon principle does 110t
apply to nonlinear systems makes generalisation difficult and study of many nonlinear
systems has to be specific for typical situations. .
! 6.2 Behaviour or Nonlinear Systems
I
i The most important feature of nonlinear systems is that nonlinear systems do not obey the
principle of superposition. Due to this reason, in contrast to the linear case, the response
of nonlinear systems to a particular test signal is no gnide to their behaviour to other Fig.6.2(a) Fig.6.2(b) Fig. 6.2 (c)
-I
I inputs. The nonlinear system response may be highly sensitive to input amplitude. For
example, a 'nonlinear system giving best response for 2 certain step input may exhibit
highly unsatisfactory behaviour when the input amplitude is changed. Hence, in a
nonlinear system, the stability is very much dependent on the input and also the initial
state.
When excited by a sinUSOidalinput of constant frequency and the amplitude is increaser
--from-IoW-vliJues,the output frcijiiency-aCCSom,q)oih[exactly matches wit}. the inpu
frequencyand continue to remain as such thereafter. This phenomenon which results in ;
synchronisation or matching of the output frequency with the input frequency is caller
frequency entrainment or synchronisation.
I Further, the nonlinear systems may exhibit limit cycles which are self-sustained
oscillations of flxed frequency and amplitude. Once the system tmjectories converge to a
limit cycle, it will continue to remain in the closed trajectory in the state space identified
as limit cycles. In many systems the liurlt cycles JI1'Ccundesirable particularly Whenthe
6.3 Methods of Analysis
Nonlinear systems are difficult to analyse and arriving at general conclusions arc tedious
I Linearization Techniques: in reality all systems are nonlinear and linear systems are
only approximations'of the nonlinear systems. In some cases, the linearization yields
I
I
useful infonnation whereas in some other cases, linearised model has to be modified
Whenthe operating point moves from one to another. Many techniqueslike perturbation
method, series approximation techniques, quasi-linearization techniquesetc. are used for
Iinearisea nonlinear system.
Phase Plane Analysio;:This method is applicabte to second order linear or nonlinear
Fignre 6.1 systems for the study of the nature of phase trajectories near the equilibriumpoints. The .-
A nonlinear system, when excited hy a sinusoidal input; may generate several hannonics system b.ehaviouris qualitatively analysed along with design of systemparameters so as
I and soft spring are as shown in Fig. 6.2(a), Fig. 6.2(b) and Fig. 6.2(c) respectively. For a
hard spring, as the input frequency is gradually increased from zero, the measured
amplitude, frequency :md'stabilfty of these.limit cycles. Accuracy is better for 'hfgber
order sYstemsasthey have better low pass characteristic. .
I 2
l
58
VTlve
LearnTng
~jtc:---
L1apunov's Method for StabDlty: The analytical solution of a nonlinear system israrely ,
possible. If a numerical solution is attempted, the question Of stability behaviqnr can nOI
be fully answered as solutions to an infinite set of initial conditions are needed: The
Russian mathematician A.M. Uapunov introduced and formalised a method which 8Jlows
one to conclude about the stability without solving the system equations. .
6.4 Classification of Nonlinearities "'-- ~...
The. nonlinearities are classified into i) Inherent nonlinearities and ii) Intentional '~"
.
nonlinearities.
The nonlinearities which are present in the components used in system due to the inherent
imperfections or properties of the system are known as inherent nonlinearities, Examples
---- :~.-.
Figure 6.4
are saturation in magnetic circuits, dead zone, back lash in gears etc. However iii some
cases introduction of nonlinearity may improve the performance of the system, make Ihe Relay: A relay is a nonlinear power amplifier which can provide large power
system more economical consuming less space and more reliable than the linear system' amplification inexpensively and is therefore deliberately introduced in control systems. J\
designed to achieve the same objective. Such nonlinearities introduced intentionally to relay controlled system can be switched abruptly between several discrete states which
improve the system performance are known as intentional nonlinearities. Examples are are usuaIlY.'lff, 'full forward and full reverse. Relay controlled systems find wide
different types of relays whiel, are very frequently used to perform various tasks. But it, applications in the c\lDtrolfield. The characteristic of an ideal relay is as shovn in figure.
should be noted that the. improvement in system performance due to nonlinearity is In practice a relay has a definite amount of dead zone ns shown. This dead zone is caused
possible only under specific operating conditions. For other. conditions, generally by the facts that relay coil requires a finite amount of current to actuate the relay. Further.
nonlinearity-degradesthe performance of the system. . since a larger coil current is needed to close the relay than the current at which the relay
drops out, the characteristic alw.aysexhibits hysteresis.
6.S Common Physical Nonlbiearities: The common examples of physical
nonlinearities are saturation, dead zone, coulomb friction, stiction, backlash, different Multivariable Nonlinearity: Some nonlinearities such as the torque-speed characteristics
types of springs, different types of relays etc. .. of a servomotor, transistor characteristics etc., are functions of more than one variable.
Such nonlinearities are called multivariable nonlinearities.
Saturation: This is the most common of all nonlinearities. All practical systems,' when
driven by sufficiently large. signals, exhibit the phenomenon of saturation due to
limitations of physical capabilities of their components. Saturation is a common
phenomenon in magnetic circuits and amplifiers.
Dead zone: Some systems do not respond to very small input signals. For a particular
range of input, the output is zero. This is called dead zone existing in a system.' The
input-output curve is shown in figure. .
_lion
Figure 6.3
Backlash: Another important nonlinearity commonly occurring in physical systems is
hysteresis in mechanical transmission such as gear trains and linkages. This nonlinearity
is somewhat different from magnetic hysteresis and is commonly referred to as backlash,
In servo systems, the gear backlash may cause sustained oscillations or chattering
phenomenon and the system may even turn unstable.foriarge backlash. . .'
.'
59 60
Chapter 7
This equation may be written in the standard form
Phase Plane Analysis
7.1 Introduction X + 2(<<.I"x + '''~x = 0
where; and (I). are the damping factor and undamped natural frequencyof the system.
Phase plane analysis is one of the earliest techniques developed for the study of second
order nonlinear system. It may be noted that in the state space formulation, the state Defining the state variables as x = x and x = X2, we get the state equation in the state
_ variable form as
variableschosen are usually the output and its derivatives. The phase plane is thus a state
.. plane Where the two state variables X, and X2 are analysed which may be the output
variable y and its derivative y. The method was first introduced by Poincare, a French
X1
Xl =.'(2
= -2{WI:'~:2-Wf;Xl
ma!hemadcian. The method is used for obtaining graphically a solution of the following
two simultaneous equations of an autonomous system. These equations may then he solved for phase variables x, and '2. The time response
riots of x" X2 for various values of damping with initial conditions can be p!otted.. Wh~n
x, = f,(x,.x2) the differential equations describing the dynamics of the system arc nonli~ear, It IS In
Xz = 12(>'1':<2) general not possible to obtain a closed form solution of x" X2. For example, if the spnng
force is nonlinear say (kjx + k"x3)the state equation rakes the form
Where x, = h(x1,,,,) and x2 = f.(x"x2) are either linear or nonlinear functions of the
state variables Xl and X2 respectively. The state plane with coordinate axes Xl and X2 is Xl = ~:z
called the phase plain. In many cases, particularly in the phase variable representation of
systems, take the form ' .
X2 - _Lx
- M -~x
M - !2x
2 1 fo." J
Xl = Xz
Solving these equations by integration is no more an easy task. In such situations, a
x, = f,(x"x,) graphical method kpown as the phase-plane method is found t? be very help~l. The
coordinate plane wiThaxes that correspond to the dependent vanable x, and X2 IS called
The plot of the state trajectoriesor phase trajectories of above said equation thus gives an
idea of the solution of the state as time t evolves without explicitly solving for the state. .' phase-plane. The curve described by rhe state point. (;;"X2) in the ph?se-plane with
The phase plane analysis is particularly suited to second order nonlinear systems with no respect to time is called a phase trajectory. A phase trajectory can be easily constructed
by graphical'techniques. ..
input or constant inputs. It can be extended to cover other inputs as well such as ramp
inputs, pulse inputs and impulse inputs. ' 7.2.1 Isoclines Ml!thod:
7.2 Phase Portraits
From the fundamental theorem of uniqueness of solutions of the state equations or
. ,
differentialequations, it can be seen that the solution of the state equation starting from an
Let the state equations for a nonlinear system be in the form
x, = flxvxll
x: = f:_ (x1'x2)
initial state in the state space is unique. This will be true if fJ(X"X2) and h(x"x2) are'
analytic. For such a system, consider the points in the state space at which the derivatives Whereboth i,(X,~'t2)and [o(x,.x,) are analytic.
of all the state variables are zero. These points are called singular points. These are in From the above equation, the slope of the trajectory is given by
fact equilibrium points of the system, If the system is placed at such a point, it, will
continue to lie there if left undisturbed. A family of phase trajectories starting from .' dx, ~ hex "x,) =M
different initial stales is called a phase portrait. As time t Increases, the phase portrait d_~, f,(x"x,)
graphically shows how the system moves in the entire state plane from the initial states in Therefore, the locus of constant slope of the trajectory is given by
the different regions. Since the solutions from each of the initial conditions are unique,
!he phase trajectories do not cross one ano!her. If the system has nonlinear elen)ents f2(x"x2) = Mf,(x,,x2)
_which are piece-wise linear, the complete state space can be divided into different regions The above equarion gives the equation to the family of isoclines. For different values of
and phase plane traJectoriesCOnstructedfor each ')f the regions separately. M, the slope of rhe trajectory, different isoclines call be drawn in the phase plane.
7.3 Phase Plane M9thod Kno",ing the value of M on a given isoclines, it is easy to draw line segments on each of
Consider the homogenous second order system with differential equations
d2x dx
, .
!hesejsocIiR<;S_,
.
Consider a simple linear system with state e_quatJOns
M at' + f dt + kx =_o
Dividing the above equations Weget the slope of the st<te trajectory in the X,-X2plalle as
.61 62
VTU~ VTU_X; ..
Lecirn~ng LearnTng
3. For typical values ofM, draw a large number of isoclines in x,-x2 plane
4, On each of the isoclines, draw small line segments with a slope M,
5, From an initial condition point, draw a trajectory following the line segments with
slopes M on each of the isoclines.
For a constant value of this slope say M, we get a set of equations Example 7.1: For the system having the transfer function .r.'~1) and a relay with dead
-1 zone as nonlinear element, draw the phase trajectory originating from the initial condition
x2=(M+l)Xl (3,0).
.'
which is a straight line in the XI-Xl plane. We can draw different lines in the XI-Xl plane
for different values of M; called isoclines. If draw sufficiently large number elf isoclines
to cover the complete state space as shown, we can see how the state tdjectories: are
moving in the state plane. Different trajectories can be drawn from different iIijtlal
conditions. A large number ..r such trajectories together form a phase portrait. A few
typical trajectories are shown in figare given below. .
The differential equation for the system is i:" + C = II where u is given by
+1 ife> 1
,< = 0 ll- 1 < " <; 1
{-1 tfe<-1
Since the input is zero, e = r -e- c = -c and the differential equation in terms of error will be
e+';=-l1
Defining the state variables as x, = e and X2 = ;, we get the state equations as
Xl=X2
'{:2:- -X2-U
We can identify three regions in the state plane depending on the values of e = x"
Phase plot' Region1:e=x.> 1
1
Figure 7.1 Here u = 1, so that the isoclines are given by x:z=~ or .x ..
.:.
=-
M-l
..
,
The Procedure for constructlon or the phase trajectories can be summarised' as
below:
For different values of M, these are a number of straight lines parallel to the x-axis .
I. For the given nonlinear differential equation, define the-state variables as X, and Xl
and obtain the state equations as
x, = /"(X"X2)
2; Determine tbe equation to the isoclines as
Region2: e= -1 <x, < 1
dx, = 1("1'x2) = M
dx, X,
63 ....
64
VT~ .'
Licirn'fng
Here u = 0, so that the isoclines are given by x2 = ~ or M = 1, which are parallel lines
having constant slope of 1. Trajectories are lines of constant slope .1. . The slope of the trajectory is given by
Region 3: e = ", <-1
dx! = -2x,- 5:<1+ 10 = M
Here u = ~1so t hat on su bsutution
" we get xt, = -~.+1
M or %2 == M+i
1
dx! X2
-r-Sx1' 10
These are also lines parallel to x - axis at a constant distance decided by the value of the :. .":z:::: M+""7. i' M + ?
slope of the trajectory M. .
'. When xj
-SX1
e
+~
O,
=0 c>r Xi =2
M+2 M.+2
ie., all the isoclines will pass through the point x, = 2, X2 = 0.
WhenM",O,
-5' 10
"'=27"1+"2
s=o WhenM =2,
1/3
1 -5 10
3 "2=4""1 +4'
-s
-3
,
WhenM=4,
-7/3 -5 10
-2 "'=5"1 +5"
WhenM = 8.
-5 10
"2 =10.11+ 10
Figure 7.2: Phase portrait ofExarnpl" 1
WhenM=2,
X2. == 00
Isoclines 4rawnfor all threbegioDs; are as shown in figure.' It l~ seen that :trajetOries
from either region I -or 2 approach 'the b"onnllazy between the regions and appro.!:]; the WhenM=4,
origin along a line at-I slope: The state-"lIt)setlleatany value ofx, between-t and:+1 as 5 10
this beinga dead zone-and-no:(urther-movement-to the origin along the x,"axis will be ~'(2- "2.:.tr; 2"
possible. This will result in a steady state error,.the maximum value of whieh-is.equal to
half the dead zone. However, the presence of dead zone ean reduce the oscillations and WhenM=~:
result in a faster response. In order that the steady state error.in the output is !'Muced, it'is '5 10'
better 10 bave as small a dead zone as possible. _ - x2 ="4%1:"4
Example 7.2: For the system having a closed loop transfer function _.. " , plo; the WhenM=10.
phase trajectory originating from the initial condition H,0). +21+5 : 5 10
x2=8""-8'
The differential equation for the system is given by
x+2X+Sx= 10
Let x = xtandi:::: x2
Then, Xl = x2
X2 = -2r2 - 5x, + 10
.. 65 66
VTU~
..... (J)
Learning
Therefore, the slope equation over a short interval is given by
dx. Xl +15
dx, =--;::-
With 1) known at any point P on the trajectory and assumed constant fora short interval,
~e can draw a short segment of the trajectory by using the trajectory slope dX2/dx, given
m the above equation. A simple geometrical construction given below can be used for
this purpose. .
1. From the initial point, calculate the value of o.
2. Draw a short arc segment through the initial point with (-0, 0) as centre, thereby
determining a new point on the trajectory.
3. Repeat the process at the new point and continue. .
Example 73: For the system described by the equation given below, construct tile ".
trajectory starting at the initial point (I, 0) using delta method.
:Y+x+x' =0
Let x= X1C1"dx = x2 then
Xl =X1
Figure 7.3
X, = -x" -Xf
The above equation can be rearranged as
The isoclines are drawn as shown in figure. The starting point of the trajectory is marked
at (-1,0). At (-1,0), the slope i~ 00, ie., the trajectory will start at an angle 90. From.the x, = -(x, +x, +x~ - x,)
next isoclines, the average slope is (8+4)/2 = 6, ie., a line segment with' a slope 6 is drawn SOthat 8 - x~ + xI - -.
(at an angle 80.5").Thesame procedure is repeated and the complete phase trajectory will
b~ obtained as shown in figure. At initial point 6 is calculated as 0 ~ 0+1-1 = O. Therefore, the initial arc is centred ai
point (0, 0). The mean value of the coordinates of the two ends of the arc is used [0
7.3.2Deha Method: calculate the next value of 0 and [he procedure is continued. By constructing the small
The delta method of constructing phase trajectoriesis applied to systems of the form arcs if!this way the complete trajectorywill be obtained as shown in figure.
x+f(x,i,t) "=0
Where f(x,x. t) may be linear or nonlinear and may even be time varying but must be
continuous and single valued.
With the help of this method, phase trajectory for any system with step or ramp or any
time varying input can be conveniently drawn. The method results in considerable time
saving when a single or a few phase trajectories are required rather than a completephase
portrait.
While applying the delta method, the above equation is first converted to the form
x + "',,[x + c5(x,.t,t)]=0
In general, c5(x,:i, t) depends upon the variables x, x and t, but for short
.. changesin these variables are negligible. Thus over a short interval, we have
x + "'n [x + 15] = 0, where IIis a constant.
Let us choose the state variables-as XI = X; x.- Xl..;,., then
Xl = CUn%:
Xz - -Wn(AI + .:)" Figure 7.4
12
"
.6.7 _
68
.'
VTU)e'
VTUg
l.ecirjnTng ....
Learnrng
oscillation with constant amplitude. This is shown in figure (i). The inside of the limit ...
7.4 Llmlt Cycles:
cycle is an unstable region.in the sense that trajectories diverge to the limit cycle, and the
Limit cycles have a distinct geometric configuration in the phase plane portrait, namely, outside is a stable region in the sense that trajectories converge to the limit cycle.
that of an isolated closed path in the phase plane. A given system may have more than A limit cycle is called an unstable one if trajectories near It diverge from this limit cycle.
one limit cycle. A limit cycle represents a steady state oscillation, to which 01' from In this case, an unstable region surrounds a stable region. If a trajectory starts within the
which all trajectories nearby will converge or diverge. In a nonlinear system, limit cycles stable region, it converges to a singular point.within the limit cycle. If a trajectory starts
describesthe amplitude and period of a self sustained oscillation. It should be pointed out in the unstable region, it diverges with time to infinity as shown in figure (ii). The inside
that not all closed curves hi the phase plane are limit cycles. A phase-plane portrait of a of an unstable limit cycle is the stable region, and the outside the unstable region.
conservative system, in which there is no damping to dissipate energy, is a continuous
family of closed curves. Cased curves -ef 1J.i.lcind are Dotlimit cycles because none of 7.5 Analysis and Classification of Singular Points:
these curves are isolated from one another. Such trajectories always occur as a Singular points are points in the state plane where "', = "'~= o. At these points the slope
continuous family, so that there are closed curves in any neighbourhood of any particular of the trajectory dx,ldx, is indeterminate. These points can also be the equilibriumpoints
closed CUrve. On the other band, limit cycles are periodic motions exhibited only by of the nonlinear system depending whether the state trajectories can reach these or not.
nonlinear non conservative systems. . Consider a linearised second order system represented by
As an example, let us consider the wen known Vander PoI's differential equation
d2"
dt' -P(l-x')dt
dx
+x=o
t!~]=.~ ~J[;~J
Using linear transformationx?7fft the equation can be transformedto caoonicalform
which describes physical situations in many nonlinear systems.
In tenus of the state variables "1 -" ~.,..l"2 - i,we obtain
[~lJ 0 AO][~']
z, = [)" z _,
i'1 = %2 Where, A, and A, are the roots of the characteristic equation of the system.
:t, = If(l-xDx, -x, The transformation given simply transforms the coordinate axes from XI-X, planeto z,-z,
-- --plane-lfaving the same origin, bui- does not'affecCthe nature of the roots of the
The figure shows the.phase trajectoriesof the system for )l> 0 and )l < O. Incase of)l > 0 characteristic equation. The phase trajectories obtained by using this transformed'state
we observe that for large values of 11,(0), the system response is damped arid the equation still carry the same information except that the trajectories may be skewed 0:'
amplitude of X,(I) decreases till the system state enters the limit cycle as shown by the stretched along the coordina.e axes. In general, the new coordinate axes will not be
outer trajectory. On the other hand, ifInitially x,(O) is small, the damping is negative, and rectangular.
hence the amplitude of lI,(t) increases till the system state enters the limit cycle as shown The solution to the state equation being given by
by the inner trajectory. When)l < 0, the trajectories.moves in the opposite directions as
shown in figure. . z,(t) = ei,.'z,(o;
z,(t) = e1,z,(U)
~2=Y
The slope'of the trajectoryin the z, - z, plane is given by
dz, _ A,Z, _ taue
dZ1 A.171
dz, _ (~)(dZl)
Z2 U1 Zl)
Unstable
limit
.'. h.(z,) - G:)lltC"l) (II" ", - t"(Zl/'h.,
cycle
Based on the nature of these eigen values and the trajectory in z, - 2, plane, the singuiar
(i) u>0 (ii) )l < 0 points are classified as follows.
Nodal POint:
Figure 7.5
A limit cycle.Is called stable if trajectories near the limit cycle, originating from outside 'Considereigen values are real, distinct and negative as shown ill figure (a). For this ease
or inside, converge 10 that limit cycle. In this case, the System exhibits a sustained the equation of the phase trajectoryfollows as z, = c(z,)".
7.0
VTu2
LearnTng
Where, k, '" O..,iA,) ::: 0 so that the trajectories
become a set of parabola as shown in figure (b) and'
the equilibrium point is called a node. In the; The slope
tj"
(a) Ei!cnvalues (b) PhIa<!transec:IOfiu.stablcfacus
.
u-jro
.. Consider a system 'with complexconiugete eigen values. The canonical form of the state
equation can be written as Centre or Vortex Point:
Figure 7.8
[~1]"=[" 0
%2
+)I>} o .: ] [~lJ
7-:)6) 42 ..
Consider now the case of complex conjugate eigen values with zero real parts,
ie., A,,:l.2= jro
,Usinglinear transformation, the equation becomes
.'
.'
,71 72
VTU.Ae VTU)e
.. ,.
LearnTng Learnmg
_ 2 10.
x2 - -Xl + O.lx, - X, - '3X2
Ot. B"]
+
,
The Jacobean matnx J
= [!h. !h.. = [-(1 +0 2x)
D;;rl 0.%1
x, 8:r a~
1
Let the state variables be ,T, = Y and x 2 = j. ""AZ = O.S( 0.1 f j{3Jj9)
The corresponding state model is The eigen values are complex with positive real part. The singular point is an unstable
focus.
Xl:: x~
Linearization around (-1.0)
X2 = -3x2 - 2x, I
At singular points. X, = x2 = 0 and x, = -3x2 - 2:<:, =0 : Ie-1,0) - [01 1 ]
0.1
Therefore: the singular point is at (0.0) The characteristic equation will be
A_[O -2 -3
1] A -1 1
.1-1 A_ 0.1 - 0-
The characteristic equation is IJJ - A I= 0 Therefore A2- O.IA 1 = 0
. I;'
I.e, 2 A+3 =0
-1 1' Or
AI. 1..2 = 1.0~ and -0.98. Since the roots are real and one negative and another positive. the
singular point is a saddle point.
AI. A2
node.
= -2, -I. Since the roots are real and negative. the type of singular point is Example 7:6:
Dder:mineih?_1tind <lhingularity for the following differential equation,
Example 7.5: For the nonlinear system having differential equation: x +.o.~~+';Zx + x2 = 0.
y- (0.1 - ~ j.2)y + y + y' = 0, find all singularities. Let the ~tate variables be ", = x and x2 = x
Defining the state variables as.v = X" Y = x2 the state equations are The corresponding state model is
Xl =X2 Xl =%2
X:z = -O.SX2 - 2r h- xi
' ..
.73 74
VTU~
Learnfng
At singular points, x, = x. = 0 and %. = -o.Sx, - 2X, -:rI =0
So thattbe singular points ". = 0 and -2x, -xi i....,x,(x] + 2) = 0 Chapter a
i.e. x,=O or-2 STABILITY ANALYSIS
The singularities are thus at (0,0) and (-2,0). 8.1Introduction:
Linearization abollt the singularities: For linear time invariant (LTI) systems, the concept of stability is simple and can be
formalised as per the following two notions:
a) A system is stable with zero input and arbitrary initial-conditions if the resulting
trajectory tends towards the equilibrium state.
b) A system is stable if with bounded input, the system output is bounded.
Linearization around (0,0), Le., substituting XI = 0 and X, = 0 In nonlinear systems, unfortunately, there is no definite correspondence between the two
notions. Th~ Iinear autonomous systems have only one equilibrium state and their
behaviout aliout the equilibrium state completely determines the qualitative behaviour in
the entire statep)ane. In nonlinear systems, on the other hand, system behaviour for small
The characteristic equation is III-A I= 0 deviations about the' ~i1ibrium point may be different from that for large deviations.
Therefore, locpl stability does not imply stability in the overall state plane and the two
I" I concepts should be considered separately. Secondly, in anonlinear system with multiple
2 -1 =0
J.+OS equilibrium states, the system trajectories may move away from one equilibrium point
and tend to other as time progresses. Thus it appears that in case of nonlinear systems,
i.e., A()..+ 0.5) + 2= 0 or)..' + 0.5)..+ 2= 0 there is no point in talking about system stability. Morr meaningful will be to talk about .".
the stability of an equilibrium point. Stability in the region' close to the equilibrium point
,1"..1. = (-V.2S 'f J1.119;
or in the neighbourhood of equilibrium point is called stability in the small. For a larger
The eigen values are complex with negative real parts. The singular point is a region around the equilibrium pomt, the stability may be referred to as stability in the
focus. large. In the extreme case, we can talk about the stability of a trajectory starting from
i.~:,:y;;;'h~..:. ..V...:An'lOp~~~'"
'~i~t~ ~~b~""~pa ....,"" t!l~i-.'~;l:r!; ,,",urn-J 5~ob41.~-i.d&ii.iiy,A ~ilJt}Jkphysical
Linearization around (-2.0l
illustration of different types of stability is shown in Fig. 8 1
I( -1.,0) = [~ -~sl
The characteristic equation will be
"1,"2= t..)9 and -1.69. Since the roots are real and one negative and another positive, the
singular point is a saddle point. . [c)
VTU~ VTU~
..
Learnrng LiarnTng
Equilibrium State: In the system of equation x
= f(""t), a state z, wherej{'i<,t) 1= 0 for 8.3 StabiJ.ity by the Method of Liapunov
all 'I' is called an equilibrium state of the system, If the system is linear time invariant,
namely j{x,t) = Ax, then there exists only one equilibriwn stale if 'A' is non-singular and Russian mathematician A.M. Liapunov has proposed a few theorem. for the study of
there exist infinitely many equilibrium states if 'A' is singular. For nonlinear sjrstem~, stability of the system. The most popular among this is called the "Second Method of
... there may be one or more equilibrium states. Any isolated equilibrium state can be
=
shifted to the origin of the coordinates, or /(0.1) 0, by a translation or coordinates:
Liapunov" 'or "Direct Method of Liapunov". This method is very general in its
formulation and can be used to study of stability of linear or nonlinear systems. The
method is .called 'direct' method as it does not involve the solution of the system
8.2 Stability Definitions: differential equations and-stability information is available without solving the equations
. .
The. .~===.'.::..:.=!:.::'=.:::::.......
stability. These are discussed below.
~"'..!,!~,
~:::::.-:.~;
~..:.:. =.!;;--=.
:.:~:-==!~. :!"..: =!'!'~--=~:
:j~~ of
.
which \s definitely an advantage for nonlinear systems.
obtained by this method is precise and involved no approximation.
The stability information
x
Stability: An equilibrium state x, of the system = lex, tj is said 10 be stable if f6r each First Method of Liapunov: The first method of Liapuoov, though rarely talked about, is ,IW
real number s c- 0 there is a real number o(e,to) > 0 such that OXo :s; (; imp)i~s. -"'.11 essentially a theorem stating the conditions under which system stability information can
1I1P(t:xq,to) -",.11 :s; 6 for all t2: to The real nnrnber S depends on & and in general, also be inferred by examining the simplified equations obtained through local linearization.
a
depends on to If does nOI depend on to, the equilibrium state is said to be uniformly This theorem is applicabi unly to aUtUnOHWUS systems.
stable.
Asym;;>!ctlc Stability in the Large: If asymptotic stability holds for all statee from A scalar function Vex) is said to be negative definite in a region n if Vex) < 0 for all
Which trajectories originate, the equilibrium state is said to be asymptotically stable in the
nonzero states x in the region nand V(O) = o.
large. An equilibrium state x, of the system :it =!(""~)is said to be asymptotically A scalar ful!ction Vex) is said 10 be positive semi-definite in a region n if it is positive for
stable in the large, if it is stable and if every solution converges 10 x, as '1' increases all states in the region n and except at Ihe origin and al ce!'.ain Other stales. where it is
~-'::"'::"~~]':;:":",~,,,, '"---] -'''~~~n :......-Ju.y:.....
...o~:.:..:~:;
;.: ~ :!'..:.. !~~
:~6 .... ~Ui. zero.
there is only one equilibrium slate in the. whole stale space. .
A scalar function Vex) is said to be negative semi-definite in a region n if it is negative
The above said definitions are re~resented graphicallyin Fig. 8.2 for all stales in the region n and except at the origin and at certain other states, where it is
zero.
/::--~'>'t::(;s 1I.(t~I< Ill(ojll;'6 A scalar function Vex) is said to be indefinite if in the re~jon n it aSS1.1rnt"~
hOlh n, ...;th,~
I ?;::/ y~II<~ and negalive values, n~ matter how small the region fl is. -
Examples:
.
x: +-'i!
\~
Vex) - - Positit?e definite.
i.e.,
Example 8.2:--xi - 3x~ + xi + .J.x,x, + 8x,x,
A necessary and sufficient condition in order that the quadratic fonnxTAr, where A is an Successive principal minors are
nxn real symmetric matrix. be negative definite is that the determinant of A be positive if
n is even and negative if n is odd. and the successive principal minors of even order be 1> 0; I~J<: 0
_' positive and the successiveprincipal minors of odd order be negative.
Hence, the given quadratic form is indefinite
i.e.,
au < c, tau
au a.:1
"22
>0
aiz "13
r- 1 -11
rQu
S1
all
a2Z a:, < 0...
au a"
IAI > 0 :f n Is ."en and IAI < 0 If IIis odd. -- l-~-2 ':1-11
Successive principal minors are
A necessary and sufficient condition in order that the quadratic.formxTAr, where A is an
nxn real symmetric matrix, be positive semi-definite is that the determinant of A be -1 < 0,
. 1-11 -3-
11-" 0
singular and the successive principal minors of the determinant of A be nonnegative.
i.e., Hence the quadratic form is negative definite,
...
all :::: la nl~1
0, a.,
ll (J
~
1\
v, laa"u
au
au
a",
a"
4:3 ~
au,
an
0....._, r"
"21'
::
anI
ai
a!.
".2
... a
...
'1n =0
1R
ann
I 8.5 Second Method of Liapunov
The second method of Liapunov is based on a generalization of the idea that if the system
,~ .:.:: '':::Y=";::':~:;:.!~)':::.!;!: ":""'l_':!:~~::';:-''Z::~~;:.. :!-.:::: :.~;: ::~:::-:~ :::::-0:" :: ~!:.:.,.-.~-.-:~-
displaced within the domain of attraction decays with increasing time until it finally
A necessary and sufficient condition in order that the quadratic form xTAr, where A is an assumes it', minimum value at the equilibrium stale, The second method of Liapunov
nxn real symmetric matrix. be negative semi-definite is that the determinant of A be consists of determination of a fictitious energy function called a Liapunov function. The
singular and all the principal minors of even order be nonnegative and those of odd orders idea of the Liapunov function is more general than that of energy and is more widely
be nOD positive. applicable, Liapunov functions are functions of Ail 1\.2. A3 , .. An. and t. \V(. ~":"ii\Jt.:,
i.e., Liapunov functions Vex,. x,.
X3 xn t) or V(X.I)or Vex) if functions do not include
...
r"
t explicitly. In the second method of Liapunov the sign behaviours of V(X.I)and its time
r
at2
a11 :So 0,JaQll a" 12: V.
a12
,,>, a", a" S u.... ..... au: Qi!:2
... ainl
a=;",
derivative b( x, r) give information on stability. asymptotic stability or instability of the
2> 2: aOl 0131
l1 a:rz a.. ...
=U equilibrium state under consideration at the origin of the state space.
On, all2 Rnn Theorem I: Suppose that a system is described by x = f(x.t). wheref(O.I) = 0 for all t.
If there exists a scalar function V(x.t) having continuous first partial derivatives and
Example 8.1: Using Sylvester's criteria. determine the sign definiteness of the following satisfying the following conditions.
quadraticforms
L V(X.I) is positive definite
10xi + xi + 4x;+ 4x,x, + 6.~,x,+ 2X2X,
2. V(x.t) is negative denllite .,.
...A=[~!3 1 4
~J I.e. IAI > 0
Then the equilibrium state at the origin is uniformly asymptoticallystable. Ifin addition
V(..,.t) ... 00 as IIxll ... cc then the eqUilibrium state at the origin is uniformly
asymptotically stable in the large.
Successiveprincipal minors are
A visuai analogy may be obtained by conSideringthe surface i' = ~x~ + ~kxi. 111isis a
10> O' 1210 21>
1 0 cup shaped surface as shown. The constant V loci are ellipses on the surface of the cup.
Let (rf. :r:) be th" initial condition. If one plots trajectory on the surface slJown.thc
Hence.the given quadraticform is positive definite. " representative point X(I) crosses the constant V curves and moves towards the lowest
poinl of \.he cup which is the equiiibrium poim.
"
80
VTU,2
I ;.:..:...:to'Ioc;>.......,
'.,.~~~.u ; n. '~
Example 8.4: Determine the stability of following system using Liapunov's method.
x1=-x -x:z
J and Xl=Xl-X~
.
Example 8.5:' Determine the stability of following system using Liapnnov's method.
x.=.t .. cm-.d:i-,,=~x.-., ..
Let us choose Vex) = xi + xi
Then vex) = 2x,x, + 2",*, = 2x,", + 2xi -x,- x,) = -2x~
r-"''''-:o!'UP co!",",!,Apfin,tp fnnrtion. Tf V{Xl is '0 he' vanish identicallv for t ~ tr. ~h~!l
'Th;r 'co.,
x~';"~~;be'z-;;rofor all!;:: to. ... x, = O. Ther. x, = =
-x, - .t, 0 i.e.x,must be Z lO.
This means that Vex) vanishes identically only at the origin. Hence the equilibrium state
at the origin is asymptoticallystable in the large.
Example 8.6: Determine the stability of following system using Liapunov's method.
Xi = Xi. and x~= -X2 - x:
Let us choose V(~) = xr + x;
Fig. 8.3 Energy function and movement of states
Then VCx) = 2x,x, + 2x,x, = 2x,x, + 2x2: -x, -xD = 2x,x, - 2xi' - 2,,:x,
~ yl'''' J - - - ;... J .-_ , , .." _._.}\ .. , _: .. ~. ~ _
This is an indefinite funr.tio~
O.. If there exists a scalar function V(x,t) having continuous first partial derivatives and
satisfyingthefollowing conditions.
I. V(x,/) is positive definite
-;-.r~...:::~ :..;,~~Z.:.:.:==:~:!:~::!~.:-
Let us choose another Vex) = xi + 2x?
This is negative semi definite function. If Vex) is to be vanish identicallyfor t;:: to.then
..
3. V(0( t; xo,to),t) does not vanish identically in t 2: 10 for any 10and any Xo 'f 0, X2must be zero for all t 210. :. X, = O. Then ;(.,= -x, -,rJ == 0 i.e'''lntltstbe zero.
where 0(t: x.,to) denotes the trajectory or solution starting from Xo at Ij). __ This means liIat 1'(,,) ,.nishes identieltlly-elliy-at-th~rigin. Hence the equilibrium state
4l i..il~Vligill is 4~)'mv\ui.t\;cUJy~idUIt;ill Ult ii1I~~.
Then the equilibrium state at the origin of the system is uniformly asymptotically stable in
the lar~e.
.. If however, there exists a positive definite scalar function V(x,t) such that ll(x,t) is
identically zero, then the system can remain in a limit cycle. The equilibrium state at the
origin, in this case, is said to be stable in the sense of Liapunov.
Example 8.7: Consider a nc.nlinearsystem governed by tlle state eqllations
Le'orn7ng-
Therefore, for asymptotic stability we require that the above condition is satisfied. The He~~e, for th~ asymptotic stability of the system of x
= Ax, it is sufficient that Q be
region of state space where this condition is not satisfied is possibly the region of posmve definite. Instead of first specifying a positive definite matrix P and examining
instability, Let us concentrate on the region of state space where this condition is whether or not Q is positive definite, it is convenient to specify a positive definite matrix
satisfied. The limiting condition for such a region is (1- 2X,X2) > O. The dividing Q ~~st and t?en examine whether or not P determined from A' P + PA = -Q 'is also
lines lie in the first and third quadrants and are rectangular hyperbolas as shown in posrtive definite. Note that P being positive definite is a necessary condition.
Figure, 8.4. In the second and fourth quadrants, the inequality is satisfied for all values of
XI and X2. Figure 8.4 shows the region of stability and possible instability. Since the Note:
choice of Liapunov function is not unique, it may be possible to choose another Liapunov
.. function for the system under consideration which yields a larger region of stabiliiy.
I. If Vex) = -x'Qx does not vanish identically along any trajectory, then Q may be
chosen to be positive semi definite.
2. In determining! whether or not there exists a positive definite Hermitian or real
symmetric matrix P, it is convenient to choose Q = I, where I is the identity matrix.
Then the elements of P are determined from A'P + PA = -/
and the matrix P is
Stable quadrant tested for positive definiteness.
[ 01 -1
.
-'~] [1'P"11 p",?
1'12J.L~11 p"J [0 1] [-1
Lp." p,; -1 -1 = 0
I
Conclusions:
,
L.__
.'
"
.83
84
~~~~.-------
VTU~
LearnTng
=-(xy+xi)
23 > 0 and det(p) > O. 'Therefore,P is positive define. Hence, the equilibrium state at
origin is asymptoticallystable in the large.
Example 8.9: Determine the stability of the equilibrium state of the following system,
Example 8.11: Determine the stability range for the gain K of the system given be~ow.
-1-fl [X']
A'P+PA=-/
-3 J x2
~
:i'] =
~;
'1",
[0 01 -10]["'.. ]+ ~)~
=k
.0 -'2 1 ~2
a
[ttl
-2
[-1+J -_1+11 W11 Pl1l
J lP21
+ [Pu
p.. p,:] [ -2+j -1-fl=[-1 0] In determining the stability range of k, we assume u = O.
3 P., P., -1 -3 J 0-1
4PIl + (1 - j)P'2 + (1+j)P'1 = I .
Let us choose Q =
[000]
a positive semi definite real symmetric matrix
(1- j)PII + 5P" + (I - j)1>22
'" 0
r
P is positive definite. Hence the origin of the system is asymptotically stable.
Let us so! VC, ATp+PA =-/
oj Ell P"][
a
P") + E"
-kl Pl~ PIZ. 1 0
-4P" - SP22 I
5P12 + P" = 0 .
=- Solving the above equations, we get
1 [23 -7J
:,P=;SO--7 11
For P to be Positive definite, it is necessary and sufficient that 1: - 2k > 0 and k > 0 or
o < k <6. Thus for 0 < k <6, the system is stable; that is, the origin is asymptotically
stable i~ the large. "..
. '
.85. 86
" ,
VTU)C
VTU~ LearnTng
LearnTng
If f~x) is negative definite. we see that Vex) is negative definite. Hence V(x)is a
Liapunov function. 'Therefore, the origin is asymptotically stahle. If Vex} = [,(x)f(x:'
tends to infinity as 11;< II _, "". then the equilibrium state is asymptotically stable in the
large.
8.7 Stability Analysis of Nonlinear SysteJJl'l
In a linear free dynamic system if the equilibrium state is locally asymptotically stable,
El'ample 8.12: Using Krasovsleii's theorem, examine the stability of the equilibrium state
x = 0 of the system given by Xi = -Xi and X, = x, - X, - x~
..
then it is asymptoticallystable in the large. In a nonlinear free dynamic system. however.
an equilibrium state can be locally asymptotically stable without being asymptotically
stable in the large. Hence implications of asymptotic stability of equilibrium states of
linear systems and those of nonlinear systems are quite different:
The Jacobian matrix is given by
If we are to examine asymptotic stability of equilibrium states of non linear systems,
stability analysis of linearised models of nonlinear systems is completely inadequate. :We
must investigate nonlinear systems without linearization. Several methods based on:the
second method of Liapunov are available for this purpose. They include Krasovskii's
method for testing sufficient conditions for asymptotic stability of nonlinear systems.
Schultz- Gibson's variable gradient method for generating Liapunov functions etc. .
KrasoyskU's Method:
lex) = F' ex) + F(x)
Consider the system defined by x = f(x). where x is an n-dimensional vector. Assume
that f(O)'" 0 and that f(x) is differentiable with respect to Xl where, I '" 1.2,3,.....n. The
Jacobian matrix F(x) for the system is
oh ah ah
a;; aX2 ax. = [-12 -2 ~ 6xd
Ok 8f2 0/.
FCr) = I)x1 I)X2 I)xn This is a negative definite matrix and hence the equilibrium state is asymptoticallystable.
, '
fT:x)f(x) - 'fr t (Xl' x: "D' ,00 a.II~1I , <G. Therefore the equilibrium state is
0/. !!b. af. asymptoticallystable in the large.
:x.. iJx., AX ..
Example 8.13: Using Krasovskii's theorem, exarnine the stability of the equilibrium state
Define F(x) = rex) + F(x), wherer:x) is the conjugate transpose of F(x). If:the
Hermitian matrix P(x) is negative definite, then the equilibrium state x = o is x=Oofthe system given by .r1 = -3x, +', ami i:, = x,- x, - xi
asymptotically stable. A Liapunov function for this system is vex; - f(x)f(x). ff in
addition /,(x)f(x) ... 00 as Bxll ... 00, then the equilibrium state is asymptotically stable
in the large.
fnlllf;
If F(x)is negative definitefor all x f. O. the determinant of P is nonzero everywhere
F(x) = [-;_J -1~ 3x~1
except at x = O. There is no other equilibrium state than x = 0 in the entire state space, rex) - F'(x) I P(x)
Since f(O)= 0, f(xl f. 0 for X f. 0, and Vex) = f'(x}i(x), is positive definite. Note ihat
/(x) = F(x)x = F(x)f(x) . .
jI
w~can obtain V as
-------_._._--. __ ._-- .
"
"
'>
L J o
(;:)1
'1'
f xix - '''-] ,., .
. : ',,~x- 'x_ = '~pu" 'x = 'f Aq UIlA!8=ISAs ~IIIJoO =x
" "leiS ';'n!,qmn~ "I{IJO ,u!l!quis ~ "UfW8l<~'="I{I S,!p[sAOSlU){SU!SO :"1'8 ;ldwBl<:!l
'~IIJe( "I{IO! "(qBls A((1r.I!10IdUlAnS! "11118wnllqmnlr.l
'00 ... IIXIISI; OCI ... ,(;1' - 'x - TX) +.F"+ 'x_) = (x)j(x;./.I
.:
'"IIJe( "III D!"Cquls AnUQ!lOIdw.{ses!
"leis wn!'4!l!n~ "I{I :UOp.t;l'U, '00 ... UXls" 00 ... ("X;" _. '1'_) + .(~x) -= (x)j(X;./.1
.oIiI' "(qels.A((lr.l!10Idw.{SB S! ~Iels WDflqmDOO"Ill QQU"I{pue XfJlew "lfO!l"P ""fleliIlD e Sf S!I!,L
. [IXZ- 'x1XZ-j =
'x'XZ- 0
[{:"- z:~!XZ- T-] + [fX- TJ = .
:J 0 'x'X1.-t- n
(x).:/ + (x) .:/ = (x)1/
. J