Comparative Study on Optimising the EKF for Speed Estimation of an Induction Motor using Simulated Annealing and Genetic Algorithm

This paper presents a comparative study for optimising a speed observer in induction motor sensorless control using a stochastic method. A new approach of optimising the performance of the Extended Kalman Filter using Simulated Annealing is compared with use of a Genetic Algorithm. Although the EKF is capable of estimating the motor states and speed simultaneously, in this case only the rotor speed is estimated and observed. The performance of speed estimation using both methods is compared with respect to various speed ranges, robustness relatively to motor parameter sensitivity and load torque condition. The optimisation techniques are illustrated through a MATLAB/Simulink implementation on a constant V/F controller under various operating conditions.