Tuning of Extended Kalman Filters for Sensorless Motion Control with Induction Motor

This work deals with the tuning of an Extended Kalman Filter for sensorless control of induction motors for electrical traction in automotive. Assuming that the parameters of the induction motor-load model are known, Genetic Algorithms are used for obtaining the system noise covariance matrix, considering the measurement noise covariance matrix equal to the identity matrix. It is shown that only stator currents have to be acquired for reaching this objective, which is easy to accomplish using Hall-effect transducers. In fact, the Genetic Algorithm minimizes, with respect to the system covariance matrix, a suitable measure of the displacement between the stator currents experimentally acquired and those estimated by the Kalman filter. The proposed method is validated by experiments.