You are on page 1of 6

A Novel Non-Fragile Single-Loop Voltage and Frequency Controller for Induction Generator Based Isolated Renewable Energy Conversion

System
J. K. Chatterjee, Member, IEEE and Priyesh J. Chauhan, Student Member, IEEE
Department of Electrical Engineering, Indian Institute of Technology Delhi, New Delhi 110016, India jkc@ee.iitd.ac.in, priyesh.chauhan@ee.iitd.ac.in
Abstract-Poor voltage and frequency regulation under source and load perturbations limit the use of self excited induction generator (SEIG) in isolated and dispersed generation, which can exploit wind/microhydro type renewable energy sources. In the present work, a Generalized Impedance Controller (GIC), which is a pulse-width-modulated voltage-source-inverter with a dc-link battery, is used to regulate both, amplitude and frequency of the SEIG terminal voltage using only PI controller in voltage feedback loop. The important aspect of the present work is to show the dependence and sensitivity of the dynamic response of both amplitude and frequency of the SEIG terminal voltage, following source and load perturbations, on the PI controller gain settings of the voltage feedback loop. For non-fragile system operation, optimum gain settings of the voltage-loop controller are selected within a stabilizing range to minimize the frequency disturbances caused by the source/load perturbations.

I.

INTRODUCTION

Limited resources and ever rising prices of the fossil fuels, as well as degradation of the environment due to their overuse have made electric power generation from the renewable energy sources more attractive. Also, the growing importance of dispersed generation schemes for small loads situated in remote areas has made electric power generation from stand alone self-excited induction generator (SEIG) using locally available wind or microhydro type energy sources popular. Moreover, SEIG is rugged, simple in construction, low cost and maintenance, having inherent short-circuit protection and has the ability to generate at variable speeds. However, due to poor regulation of terminal voltage and frequency of the generator [1], the need for real time intelligent control of active and reactive power at the SEIG terminals has been gaining greater concern among the researchers. Rapid growth of power electronics and microcontroller technology has allowed tremendous advancement in real-time control of the SEIG output voltage and frequency using the principles of closed loop control. The control techniques suggested till date, largely confine to only voltage regulation [2, 3] and a few discuss simultaneous control of both the parameters [4, 5]. However, these techniques lack the reliability and simplicity of the control structure. Along with simplicity and accuracy of the controller, the quality of power supply is very important in applications like a military post, a remote industry or a microwave station, where critical loads such as computers, variable speed drives, power supplies, etc. are sensitive to changes in frequency and voltage.

To resolve some of the issues mentioned above, a generalized impedance controller (GIC) operating on the principle of pulse-width-modulated voltage-source-inverter (PWM-VSI) with a dc-link battery, which is capable [6] of providing bidirectional controlled flow of active and reactive power, has recently been proposed for regulating the amplitude and frequency of the SEIG terminal voltage [7]. This has been accomplished by the GIC, using only voltage feedback loop, without using frequency feedback. Here, the gain settings of the voltage loop PI controller have been suggested for effective voltage control. In the present work, a non-fragile single voltage feedbackloop control strategy for regulation of both, voltage and frequency in a SEIG based isolated wind/microhydro power generation supplying balanced 3-phase load has been investigated, under source and load perturbation conditions. This aspect of investigation has not been attempted so far. An attempt is made here to analyze the effects of varying the gain settings of the voltage loop PI controller on the dynamic response of the amplitude and frequency of the SEIG terminal voltage, by observing their peak over/undershoots and settling time, following source and load perturbations. Moreover, the frequency disturbances caused by the source/load perturbations are minimized by optimally selecting the gain settings of the voltage PI controller within a stabilizing range [8, 9], so as to keep the system non-fragile. Though, a trade-off has to be made between the performances of voltage and frequency control, while selecting the controller gains. II. SYSTEM DESCRIPTION AND MODELING Fig.1 shows the schematic diagram of a closed-loop, SEIG based isolated renewable energy conversion system. The integrated system consists of a variable speed turbine driven 3-phase Y-connected SEIG with a 3-phase bank of excitation capacitors, GIC and load connected at the point of common coupling (PCC). The GIC is an IGBT-based 3-phase PWM-VSI with a bank of battery connected at its dc bus and a coupling reactor (Lf, rf) connected at its ac terminals. The GIC offers variable controlled impedance across the SEIG terminals according to the value of modulation index m of the inverter and relative phase angle between the fundamental component of inverter output voltage vpwm and the SEIG terminal voltage vig [6].

978-1-4244-5226-2/10/$26.00 2010 IEEE

297

For the load of resistive nature, current can be derived from the ratio of respective voltage and resistance, whereas, the load of capacitive nature can be accounted for by adding its capacitance to the excitation capacitance. B. Modeling of GIC The GIC output voltage vpwm, emulated through IGBT based PWM inverter, is derived as, Vpwm = (Vdc / 3) [2 1 1; 1 2 1; 1 1 2] [Sa; Sb; Sc], (6) where, Vpwm = [vpwma; vpwmb; vpwmc]; Vdc is the dc link voltage; and the switching functions Sa , Sb and Sc (1=ON, 0=OFF) are determined by the modulation index. The expression for the current iGIC injected into the GIC can be derived from the PCC voltage vig, the GIC voltage and the value of coupling reactance, as follows. Lf pIGIC = Vpwm Vig rf IGIC , (7) where, IGIC = [iGICa; iGICb; iGICc] and Vig = [viga; vigb; vigc]. C. Modeling of Control Scheme

Fig. 1. Schematic diagram of SEIG with load and GIC connected at PCC

The control scheme includes a voltage feedback loop to sense the amplitude of the SEIG terminal voltage Vm, which is compared with its reference value Vm*. The error is processed by a PI controller, which sets the m for the GIC inverter. The frequency f of the SEIG is set by the GIC fundamental set frequency [6] by a phase lock mechanism, which is controlled by the frequency reference *. The entire closed loop system model, having the SEIG, the GIC and the control scheme as the major components, is presented below. The load and excitation system models are integrated with the SEIG model.

The purpose of the control scheme is to regulate the SEIG terminal voltage and frequency under source and/or load perturbations. For voltage regulation, the GIC modulation A. Modeling of SEIG index m, which is responsible for setting the SEIG terminal The q-d axes stationary reference frame model of the voltage, is controlled using a PI controller. The input to the PI th induction machine [10] is used for analysis. Following controller is the voltage error given at the k sample as, * differential equations are derived [7] using the q-d Verr(k) = Vm (k) Vm(k), (8) components of the stator and rotor currents as state variables. where, Vm , the amplitude of the balanced sinusoidal SEIG voltage per phase, can be calculated as, pisq = ( Lrrsisq rM2isd + Mrrirq rMLrird + LrVsq ); 2 (9) Vm = ( (2/3) ( viga2 + vigb2 + vigc2) )1/2. pisd = ( Lrrsisd + rM isq + Mrrird + rMLrirq + LrVsd ); th Output of the PI controller, m is expressed at the k sample as, pirq = ( Lsrrirq + rMLsisd + Mrsisq + rLsLrird MVsq ); m(k) = m(k1) + kiVerr(k) + kp( Verr(k) Verr(k1) ), (10) pird = ( Lsrrird rMLsisq + Mrsisd rLsLrirq MVsd ); 2 Ls = Lls + M; Lr = Llr + M; = 1 / ( LsLr M ). (1) where, kp and ki are proportional and integral gain constants. The PI controller output m is multiplied by a unit amplitude The electromagnetic torque Te can be computed as, sine wave generated at a set reference frequency * (100 (2) Te = (3/2) (P/2) M ( isqird isdirq ), rad/sec) to create the modulating wave, which is compared where, subscripts q and d are for quadrature and direct axes; with a fixed frequency f (1.8kHz) triangular carrier wave to c subscripts s and r for stator and rotor variables; subscript l for generate the PWM switching states S , S and S for the a b c leakage component; v and i for instantaneous voltage and inverter IGBTs. current; im for magnetizing current; M for magnetizing For SEIG frequency regulation, no feedback control loop inductance; r for resistance; L for inductance; P for number of has been used. An inherently created transient as a result poles; p for time differentiation and r for electrical rotor speed. of the change in the SEIG terminal frequency with respect to The magnetizing inductance is a nonlinear function of the the GIC terminal frequency under source/load perturbations instantaneous value of magnetizing current, which is obtained causes adequate active power flow between the GIC and the by performing synchronous speed test on induction machine SEIG to restore the SEIG terminal frequency to the GIC set (see Appendix). The magnetizing current is given by, frequency. This phenomenon and the operation of the GIC are im = ( (isd + ird)2 + (isq + irq)2 )1/2 . (3) well explained in [6, 7]. The q- and d-axis voltage state equations for the excitation III. CONTROL SYSTEM DESIGN PROCEDURE system are derived as, In order to reduce the frequency deviation due to pvsq = icq / Ceq ; pvsd = icd / Ced , (4) perturbation in the active power load connected to a two-area where, Ceq and Ced are the excitation capacitor values along power system, the application of a 12-pulse line commutated the q- and d-axis respectively. converter based magnetic energy storage unit has been The q- and d-axis current state equations for the balanced investigated [11]. The unit is capable of providing R-L load are derived as, bidirectional active and lagging reactive power to the power piRLq = ( vsq RLiRLq )/LL ; piRLd = ( vsd RLiRLd )/LL, (5) system. A frequency feedback loop with optimal proportional where, RL is load resistance and LL is load inductance. control had been used to improve the transient response.

298

In a similar context, in the present work, it is proposed to study the effects of varying the voltage loop PI gain settings on the transient response of the frequency in a GIC based control scheme, where, the regulation of amplitude and Fig. 2. Feedback system for SEIG terminal voltage control frequency of the SEIG terminal voltage could be 0.9 accomplished by single, PI control based, voltage feedback 0.85 loop only, without having a frequency feedback loop. The 0.8 360 GIC provided controlled bidirectional flow of both, the active and reactive power. As the transient response of the system 340 frequency depends on the dynamic behavior of the voltage 320 1.9 2 sec 2.2 2.3 loop response, which is affected by the gain settings of the Fig. 3. Evaluation of open loop steady state system gain k voltage loop PI controller. 0.0049 ku=0.0049 In view of the above, the effects of varying the gain 0.0048 228 settings of the voltage loop PI controller, on the dynamic 227 response of the amplitude and frequency of the SEIG terminal 226 voltage, are analyzed by observing their peak tu=0.08sec 225 over/undershoots and settling time, following source and load 4.8 5 5.4 5.6 sec perturbations. For this purpose, the PI controller gain settings Fig. 4. Evaluation of ultimate gain ku and ultimate period tu are chosen from a stabilizing range in order to keep the D = tu [ arctan{ (k2ku21) } ] / 2. (14) system non-fragile. The procedure [8, 9] for designing the PI controller and evaluating the stabilizing range of its gain T = tu { (k2ku21) } / 2. (15) settings is presented next. The values of parameters of the plant model in (12), The system is considered as a typical single-input single- obtained from (13-15), Fig. 3 and Fig. 4, are given in Table I. output feedback control system as shown in Fig. 2, where Vm* TABLE I CALCULATED VALUES OF PARAMETERS OF PLANT IN (12) is the command input, Vm is the controlled output of the plant, Parameter Value and C(s) is the PI controller represented as, Open Loop Steady State Gain k 291 Apparent Time Delay D sec 0.03 (11) C(s) = kp + (ki / s).
SEIG Peak Volt/ph Vm (V) kp

Modulation SEIG Peak Volt/ph Vm (V) Index 'm'

The plant to be controlled, which includes the entire generating system except the feedback element, the comparator B. Evaluation of Stabilizing Range of PI Controller Gains An analytical characterization of the set of PID controller and the PI controller; is characterized by the transfer function, Ds parameters (kp, ki and kd), for which the closed-loop system in G(s) = [k / (1 + Ts)] e , (12) Fig. 2 is stable, is provided by the theorem suggested in [8]. where, k is the steady-state gain, D is the apparent time delay, According to this theorem, the range of kp values for which, a and T is the apparent time constant. Evaluation of these given open-loop stable plant (12), having the parameters k, D parameters is presented in the following subsection. and T as in Table I, continues to have closed-loop stability A. Evaluation of Plant Parameters with a PID controller in the loop, is estimated as, 0 < kp < kupp (= 0.004979). (16) The steady state gain k is found by applying a step change For selected discrete samples of kp within above range, the in the input m (modulation index) to the plant (SEIG and GIC) and observing the steady state output Vm (amplitude of stabilizing regions in (ki, kd) space are obtained by plotting the SEIG terminal voltage). As shown in Fig. 3, a positive the trapezoidal, triangle and quadrilateral cross sections in the (0< k p<1/ k), (=1/k) and (1/ k< kp <k upp ) step change in m from 0.8125 to 0.9125 applied at 2sec yields sub-regions respectively [8] as shown in Fig. 5. a steady-state rise in Vm from 325.3 V to 354.4 V. 5e-3 The value of k is obtained from the ratio of change in 0<kp<(1/k) kp=(1/k) output Vm to change in input m as, The parameters D and T in the model (12) can be determined by measuring the ultimate gain and the ultimate period of the plant using the closed-loop configuration shown in Fig. 2. While keeping k i =0, the controller gain kp is adjusted until the system becomes marginally stable and its controlled output Vm begins oscillating continually at constant amplitude as shown in Fig. 4. This value of controller gain is called the ultimate gain ku and the time period for one oscillation of the controlled output is called the ultimate period tu. From these values, D and T are calculated [9] as,
Proportional Gain kp

Apparent Time Constant T sec

0.01294

k = (Vm Vm) / (m m).

(13)

(1/k)<kp<kupp

1e-3 5e-5 Derivative Gain kd -5e-5 0 0.1 Integral Gain ki 0.2

Fig. 5. Stabilizing regions of (ki, kd) for 0<kp<kupp

299

kp 0.00100 0.00140 0.00180 0.00220 0.00260 0.00300

ki 0.185 0.184 0.183 0.182 0.181 0.177

Combination Name PI-1 PI-2 PI-3 PI-4 PI-5 PI-6

kp 0.00344 0.00375 0.00405 0.00435 0.00465 0.00495

ki 0.170 0.164 0.155 0.145 0.129 0.098

Combination Name PI-7 PI-8 PI-9 PI-10 PI-11 PI-12

SEIG frequency SEIG Peak Volt/ph Vm (V) f (Hz)

TABLE II PI CONTROLLER GAINS SELECTED FROM STABLE REGIONS FOR ANALYSIS

400

200

0 54 52 50 48 0

GIC 'ON' SPEED CHANGE (Type-b & c) 399 ohm/ph ON

Kp-Ki CHANGED or PERTURBATION REMOVED PERTURBATION APPLIED

10

sec

20

Fig. 6. SEIG terminal voltage peak Vm and frequency f ; events time

In the present system, a PI controller has been used, therefore, the parameter kd has been ignored. Table II lists the discrete combinations of the PI controller gains along with their designated names, which are selected for analysis from their non-fragile, stabilizing regions shown in Fig. 5. IV. RESULTS AND DISCUSSION

The integrated generating system shown in Fig. 1 has been modeled in MATLAB using Simulink toolbox and solved using an implementation of multistep Runge-Kutta method in ode23tb solver. The induction machine parameters are presented in the Appendix. The Simulink program begins with the prime-mover speed set at 164.1 rad/s, which results in a small stator voltage due to residual magnetism. The moment, when a bank of wyeconnected 128.3-F excitation capacitors is inserted across the machine terminals, is considered as reference time t=0sec. The SEIG terminal voltage attains amplitude of 334.8V at 50.68Hz frequency in nearly 0.3sec following the voltage build up process. The frequency is measured using a phaselocked loop. At 0.5sec, a resistive load of 399ohm/ph is switched on, which brings the amplitude and frequency of the SEIG terminal voltage (PCC voltage) at their reference values of 325V and 50Hz respectively. The GIC is synchronized with the PCC at 1.0sec with the amplitude and frequency of the fundamental component of the GIC output voltage at 325V and 50Hz respectively, thus, the GIC floats across the PCC with no power exchange. The SEIG delivers the active and leading reactive power demanded by the load and the excitation capacitors respectively. In the present investigation, the transient behavior of the magnitude and frequency of the SEIG terminal voltage has been studied under the following perturbations: Type-a: rise in the generator shaft speed from 164.1rad/sec to 174.9rad/sec; Type-b: insertion of R load 54ohm/ph; and Type-c: insertion of 0.8pf lag RL load 54ohm, 162mH. In order to examine the integrated SEIG-GIC closed-loop system operations for all the combinations of (kp, ki) listed in Table II, the events illustrated in Fig. 6 are arranged in following manner for the perturbations of type-a: 1) Change in the (kp, ki) combination at every xth sec, where, x=3, 5, 7, , 25; 2) Application of perturbation at every (x+1)sec followed by removal of the same at every (x+2)sec; For the perturbations of type-b and type-c, above pattern is repeated, except the shaft speed is raised at 1.5sec from 164.1rad/sec to 174.9rad/sec and held till end.

The simulated transient waveforms of the SEIG peak terminal voltage and frequency are shown in Fig. 7-9 for the perturbations of type a, b & c respectively. The waveforms of the SEIG peak terminal voltage are shown separately for each combination of the controller gain setting, while those of frequency are shown overlapped along with insets magnifying the same waveforms to get a closer view of their peak undershoots, overshoots and settling instants. The rise in the generator shaft speed results in a momentary dip in the peak SEIG terminal voltage due to excess active power flow from generator, causing larger stator current, which in turn results in higher stator voltage drop. The closed loop control action sets the m to such a value, so as to enable the GIC to provide adequate reactive power, in order to restore the peak SEIG terminal voltage back to its reference value. During the same period, availability of excess shaft power leads to momentary rise in the SEIG terminal frequency, allowing the growth of in negative direction to such a value, so as to enable the GIC to absorb adequate active power from the PCC, in order to restore the SEIG terminal frequency back to the GIC set reference frequency. Insertion of unity power factor (upf) or non-upf load at the SEIG terminals results in a momentary fall in the amplitude and frequency of the SEIG terminal voltage due to excess load. The restoration of both the parameters is made by a control action similar to that explained for rise in shaft speed, except the growth of being in positive direction in this case, so as to enable the GIC to provide adequate active power to the PCC. For all the three types of perturbations, it has been observed from the peak SEIG terminal voltage waveforms, that the rise/fall time values are lower for PI-1 to PI-4. The values of settling time, however, are observed to be larger on account of more oscillations of the under-damped system. For PI-5 to PI-8 the peak SEIG terminal voltage waveforms show the response like that of a critically-damped system with smaller values of settling time. The least time taken by the peak SEIG terminal voltage to settle within an error band of 0.1% has been found to be 169ms, 139ms and 110ms for PI-5, PI-6 and PI-7 respectively, for the perturbations of type a, b and c. For PI-9 to PI-12, the values of settling time are noticed to be increasing as the system becomes over-damped. The values of % peak over and undershoot of the SEIG terminal voltage amplitude are found to be decreasing from PI-1 to PI-12 for all kind of perturbations. For the perturbations of types a, b and c respectively, 2.93%, 3.27% and 0.03% are the lowest peak overshoots and 3.71%, 8.43% and 6.8% are the lowest peak undershoots. These are occurring for PI-12.

300

340 PI-1 330 320 310 4 340 SEIG Peak Voltage/ph Vm (V) PI-4 330 SEIG Frequency f (Hz) 320 310 10 340 PI-7 330 320 310 16 340 PI-10 330 1718 PI-11 19 20 PI-12 21 11 12 PI-8 1314 PI-9 15 5 6 PI-5 7 8 PI-6 9 PI-2 PI-3

50.8 50.79 50.78 49.83 50.77 50.4 50.76 50.75 50.2 50.74 49.82 49.81 49.8 49.79 49.78 49.77 49.85 49.84 50.6 PI-1 PI-2 PI-3 PI-4 PI-5 PI-6 PI-7 PI-8 PI-9 PI-10 PI-11 PI-12

50 50.1 50.05 49.8 50 49.95

320 49.6 310 22 sec 23 24 sec 25 26 sec 27 (x+1)

49.9 (x+1.2) sec (x+1.4) (x+2)

Fig. 7(a). SEIG peak voltage/ph, under shaft speed perturbations for PI-1to12
340

Fig. 7(b). SEIG frequency, under shaft speed perturbations for PI-1to12
50.4 50.02 50.015 50.01 50.2 50.005 50 (x+1.2) SEIG Frequency f (Hz) 50 49.22 49.8 49.21 49.2 49.19 49.6 49.18 50.16 49.17 50.18 50.22 50.2 (x+1.3) (x+1.4)

PI-1 300 4 340 SEIG Peak Voltage/ph Vm (V) 5 6

PI-2 7 8

PI-3 9

PI-4 10 11 12

PI-5 13 14

PI-6 15

300

340

PI-7 16 340 17 18

PI-8 19 20

PI-9 21

300 49.4

49.16 49.15 49.14

50.14 50.12 50.1 sec (x+2)

PI-10 300 22 sec 23 24 sec

PI-11 25 26 sec

PI-12 27

49.2 (x+1)

49.13

Fig. 8(a). SEIG peak voltage/ph, under R-load perturbations for PI-1to12
330 315 PI-1 300 4 330 SEIG Peak Voltage/ph Vm (V) 315 PI-4 300 10 330 315 PI-7 300 16 330 315 PI-10 300 22 sec 23 24 sec PI-11 25 26 sec PI-12 27 17 18 PI-8 19 20 PI-9 21 11 12 PI-5 13 14 PI-6 15 SEIG Frequency f (Hz) 5 6 7 8 9 PI-2 PI-3

Fig. 8(b). SEIG frequency, under R-load perturbations for PI-1to12


50.4 50.03 50.02 50.2 50.01 50 (x+1.2) 50 49.34 49.335 49.8 49.33 49.325 49.32 49.6 49.315 49.31 49.305 49.4 49.3 49.295 49.2 50.12 50.14 50.18 50.16 50.2 50.22 (x+1.3)

(x+1)

sec

(x+2)

Fig. 9(a). SEIG peak voltage/ph, under RL-load perturbations for PI-1to12

Fig. 9(b). SEIG frequency, under RL-load perturbations for PI-1to12

The response of the SEIG terminal frequency does not follow similar trends, as that for the response of the peak SEIG terminal voltage, for similar types of perturbations. From PI-1 to PI-12, the settling time of the SEIG terminal frequency decreases initially, attains the least value and increases later for type a perturbations; it remains almost

same for type b perturbations; and it increases for type c perturbations. The least time taken by the SEIG terminal frequency to settle within an error band of 0.2% has been found to be 173ms, 202ms and 165ms for PI-7, PI-2 and PI-1 respectively for the perturbations of types a, b and c. The values of % peak over/undershoot of the peak SEIG terminal

301

voltage for above PI combinations are observed to be near maximum for corresponding types of perturbations. Thus, higher values of over/undershoots of the peak SEIG terminal voltage assist the SEIG terminal frequency to settle at a faster rate. The values of % peak over and undershoot of the SEIG terminal frequency are found to be decreasing from initial to final PI combinations for the perturbations of type a and c; whereas, for the perturbations of type b, the values of % peak overshoot increases and % peak undershoot decreases. For the perturbations of type a, b and c, 1.5%, 0.25% and 0.29% are the minimum values of the peak overshoots respectively for PI-12, PI-1 and PI-12 and 0.33%, 1.64% and 1.35% are the minimum values of the peak undershoots respectively for PI-10, PI-12 and PI-12. In order to compare the transient responses of the amplitude and frequency of the SEIG terminal voltage under the perturbations, and for the combinations of the PI gain settings under consideration, an analytical characterization of these transient responses has been done using the integral square error criterion. The performance index C is defined as,

V.

CONCLUSION

In the present investigation, in an SEIG-GIC system operating with single voltage-PI-control-loop, having no frequency feedback, the gain settings of PI controller can regulate dynamic response of both voltage and frequency under source/load perturbations. In order to keep the system non-fragile, the PI gain settings have been selected from their stabilizing range, which has been evaluated for the given system. Combinations of the gain settings of the voltage PI controller have been suggested for optimizing the control performance of both amplitude and frequency of the SEIG terminal voltage. For the considered load and source perturbations, the optimum gain settings of the PI controller are identified, which yield the best performance index for optimized transient response of both the amplitude and frequency of the SEIG terminal voltage. APPENDIX

Induction Machine Rating: 5hp, 475V, Y-Connected, 4 Pole; Induction Machine Parameters: rs = 3 , rr = 5.5 , Lls = Llr = 5.1mH; C = 0- ( wvm Vm2 + wf f 2 ) dt, (17) Induction Machine Magnetizing Characteristics: where, Vm = Vm Vm* and f = f f * are the deviations of M = 0.1956 0.122im + 0.039im2 0.0025im3; amplitude and frequency of the SEIG terminal voltage Coupling Reactor Parameters: respectively from their reference values; wvm and wf are rf = 1.68 , Lf = 64.94 mH; the corresponding weighting factors, chosen as 0.1 and 3.6 REFERENCES to obtain the optimum control performance for both, the voltage amplitude and the frequency under all types of [1] Lai, L. L. and Chan, T. F., Distributed Generation - Induction and Permanent Magnet Generators. John Wiley and Sons, 2007. perturbations. [2] Ahmed, T., Nishida, K., Nakaoka, M., Lee, H.W., "Self-excited Fig. 10 shows the performance index values for the induction generator with simple voltage regulation scheme for wind combinations of PI gain settings under the perturbations of energy," in Proc. Industrial Electronics Conf. IECON04, vol.1, pp. 8691, Nov. 2004. type a, b and c for optimum control performance of both, the [3] Ammasaigounden, N. and Subbiah, M., "Microprocessor-based voltage amplitude and frequency of the SEIG terminal voltage. PI-10 controller for wind-driven induction generators," IEEE Trans. gives the best performance index for type a and c Industrial Electronics, vol.37, no.6, pp.531-537, Dec. 1990. perturbations, whereas, for type b perturbations, the [4] Marra, E. G. and Pomilio, J. A., "Induction-generator based system providing regulated voltage with constant frequency," IEEE Trans. performance index improves monotonically from PI-1 to PIIndustrial Electronics, vol.47, no.4, pp.908-914, Aug. 2000. 12. From the above, PI-10 is identified as the optimum [5] Singh, B., Murthy, S. S., Gupta, S., "An improved electronic load controller for SEIG in micro-hydel applications," in Proc. Industrial controller setting, which yields the best performance index Electronics Conf. IECON03, vol.3, pp. 2741-2746, Nov. 2003. for optimized transient response of the amplitude and [6] Chatterjee, J. K., Perumal, B. V. and Gopu , N. R., Analysis of frequency of the SEIG terminal voltage under all perturbation operation of a self-excited induction generator with Generalized Impedance Controller, IEEE Trans. Energy Conversion, vol.22, conditions in view.
0.95 Performance Index 0.85 0.75 0.65 1 2 3 4 5 6 7 8 PI Combination 9 10 11 12 Vm-&-f Optimum Control Type a Type b Type c no.2, pp.307-315, June 2007. Perumal, B. V. and Chatterjee, J. K., Voltage and frequency control of a standalone brushless wind electric generation using Generalized Impedance Controller, IEEE Trans. Energy Conversion, vol.23, no.2, pp. 632- 641 June 2008. [8] Silva, G. J., Datta, A. and Bhattacharyya, S. P., New results on the synthesis of PID controllers, IEEE Trans. Automatic Control, vol. 47, pp. 241-252, Feb. 2002. [9] Silva, G. J., Datta, A. and Bhattacharyya, S. P., PID tuning revisited: guaranteed stability and non-fragility, in Proc. American Control Conf., Anchorage, AK, pp. 5000-5006, May 2002. [10] P. C. Krause, O. Wesynczuk, and S. D. Sudhoff, Analysis of Electrical Machinery. New York: Wiley, 2002. [11] Banerjee, S., Chatterjee, J. K. and Tripathy, S. C., Application of magnetic energy storage as load-frequency stabilizer, IEEE Trans. Energy Conversion, vol.5, no.1, pp. 46-51, March 1990. [7]

Fig. 10. Performance index vs combinations of PI gain settings

302

You might also like