Robust speed sensorless induction motor drives

Abstract

Induction motors are widely used in industry due to the fact that they are relatively cheap, rugged and maintenance free. As a consequence, much attention has been given to the motor torque and speed control. The control schemes available today require information regarding speed of the motor, which can either be obtained by using speed sensors or without speed sensors. Speed sensors have several disadvantages from the standpoint of drive cost, reliability, inertia and noise immunity. Advantages of eliminating speed sensors thus have been a strong motivation to develop speed sensor less induction motor drives for industrial drives. Several control strategies of sensor less control are available in literature. This paper is an attempt to explore the possibility of estimation of rotor speed with the help of extended Kalman filter trained recurrent Neural Network. The speed estimation is made robust by simultaneously adapting the rotor resistance and rotor flux which are also done by the same Neural Network. The training is very fast as it requires only one iteration. The proposed scheme is studied on an induction motor and it gives better performance as compared to the already existing algorithms in the literature.