The Comparisons of Optimized Extended Kalman Filters for Speed-Sensorless Control of Induction Motors

Abstract

This paper presents the comparisons of optimized extended Kalman filters (EKFs) using different fitness functions for speed-sensorless vector control of induction motors (IMs). In order to achieve high performance estimations of states/parameter by EKF algorithm, state and noise covariance matrices must be accurately selected. For this aim, instead of using time-consuming trial-and-error method to determine those covariance matrices, in this paper EKF algorithm is optimized by differential evolution algorithm (DEA) and multi-objective DEA (MODEA) with the utilization of different fitness functions. The optimally obtained set of each covariance matrices is used in EKF algorithm built on the same IM model and thus, the estimation results of the optimized EKF algorithms are compared in real-time experiments in order to conclude which fitness function is better for motion control applications.