ISSN: 1990-7958 © Medwell Journals, 2008 International Journal of Electrical and Power Engineering 2 (3): 179-184, 2008 A Fuzzy IRFOC Application Based Speed Sensorless Control of Induction Machine Using a Speed and Load Torque Observer K, Abed, K. Nabti and H. Benalla Department of Electrical Engineering, Faculty of Engineering Sciences, ‘Mentouri University Route d’Ain El Bey, Constantine, Algeria ‘Abstracts In dis sady, a Fuzzy controler of an Tadiect Rotor Feld Orenied Conrol (FOC) wing a speed observer for an Induction Machine (IM) dive, feedthrough an ameliorate voltage inverter is proposed. A superiority ofthe proposed fizzy controller over conventional PI controller n handling nonlinear sch 3s an induction motor has been effectively demonstrated by comparing FT speed controller with the conventional PI under varying operating conditions like step change inspoad reference and torque referee, Observers se a model to predict the speed by the phase currents and voltages as state variables, these estimates are improved ty an error feedback compensator that measures the difference between the estimated and acta values. The observed value of speed i then wsed by the FOC to adjust the PWM waveform in exactly the same way 38 an actual measured value Key words: IRFOC, sensoless contol, Induction Machine (IM), ameliorate tree level voltage inverter, PWM, Fuzzy-PL(F-PD) contoler, Luenberger observer, simolink INTRODUCTION Tndvetion motors are widely used in industrial applications because they are less cost, more rugged and roliable than DC motors. Though induetion motors have a few advantageous characteristics, they possess nonlinear and time-varying dynamio interactions. Using conventional PI controller, it is very difficult and complex. to design @ high performance induction motor drive ‘The fuzzy logic controller is attractive approach, which can accommodate the motor parameiric variations and difficulty in obtaining an accurate mathematical model of induction motor due to rotor parameter and load time constant variations, Tn order to have fast transient response, the controller must have the robustness against speed variations and extemal perturbations. The fuzzy logic is applied to optimize the PI conteoller gains which designed to optimize the step response of the system, ‘In this study, we describe the speed control strategies of an IRFOC for IM. Next, with help of the Matlaby Simulink, we propose the F-PI controller, which is suitable for speed control without mechanical sensors, an observer is used for predict the load torque and speed, Finally, we compare the F-PI simulation results to conventional proportional-plus integral controller. The simulation results validate the robustness and reliable of the proposed F-PI controller for high performance of inchiction motor dive, ‘THREE LEVEL VOLTAGE INVERTER As for the two level voliage inverters, we ean produce a three-phase three level voltage inverter in complete bridge by assembling three hal-bridges using common capacitive divider (Fig. 1). Fig. 1: Three level newtral-point-clamped inverter cireuit Corresponding Author: K. Abed, Department of Flectrical Enginesring, Faculty of Engineering Seiences, Mentouri University Route d’Ain El Bey, Constantine, Algeriaser Eng, 2 (3): 179-184, 2008 a Fine eno eho 8) atten ue eo ako ess nies tein eas desta eee T ‘ete ame inatee}H— 7 aa Sz, ko. is we Fig 3 Fieki oriented controler lock — diagram (Abed ra, 20056) fg) fa Foxe &t ‘ Fig. 2 Prinpel of sawtoothsimsoidel command wih | ie] iwocaiers oe is hg ss Diet PACA cnt tit |S : vollage inverter are given in Abed ef al. (2006) we use the Savrtooth-Simusoidal command with two carriers This strategy is defined by two identical carriers, which cone is delay compared to the other by @ tull-period (Fig. 2. FIELD ORIENTED CONTROL STRUCTURE A block diagram for an IRFOC can be seen on Fig. 3. This design uses a more robust structure fnown as Indirect Rotor Field Oriented Control, meaning that the rotor angle isn’t determined directly by measuring the air gap flux with halleffact sensors. These sensors are not particularly suited for use in ergo industrial motors as they oan be fragile and sensitive fo temperature change (Abed ef al, 2006 bs; Rajashekara eral, 1986). ‘The dynamic model of an induction motor ean be represented according to usual d-q axes components in asynchronous rotating frame as follows Where. M L, LL" R, ‘The decoupling control between d ane q axes ean be achieved by aligning the roter flux veetor to the d-axis and sting the rotor fTus linkage to be constant, which means 4, . © at rated flu 180Int. J. Elec, Power Eng, Substituting (2) in (1) yields The elecvomagnetio torque equation and the mechanical speed motor are elated by: im « a Where, the electromagnetic torque equation is T=x, Moi 6 7, ie The decoupling contol system is given by: (i tents 7 Where: o DESIGN OF F-PI CONTROLLER FOR INDUCTION MOTOR ‘The purpose of this study, is 10 synthesize a controller without the exact knowledge of a model, ‘numerically simple and simulated on Matlab/Sirwulink allowing good performance in terms of overshoot and fast and accuracy under the speed and load variations, The Fuzzy Logie Toolbox bused controller architecture is shown in Fig, 4 ‘The fuzzy logie controller employs speed eeror and change of speed error as inputs, the reference torque is output. 181 J: 179-184, 2008 I~ rorey | 7 ‘Gime or a Fig. 4: Fuzzy controller architecture by wuns Ps OM » as Za 55h TE St (Ere x P i as Ade dea a2 0 a2 aa os aR (©) Chang of ere s M L os ° ‘0s ass ou ow Fig, 5: Membership function of input, output variables ed)= NN o ‘et = e(8) ef) o And ituses following linguistic labels: {NL (Negative Large),.NM (Negative Medium), NS (Negative Shor), 2 {Zero}, PS Poaitive Shor), PM (Positive Medium), PL Positive Targe)}. Each fuzzy label has an associated membership furetion. The membership functions as shown in ig. 5 ‘The control rules are represented as a set if then rules, The fuzzy rules of proposed controller for speed conizol of induction motor are presented in Table 1. And formulated a follows fe (kis NL and Ae (k) 8 N then Te*() is ZE.