A Robust EKF Based Speed Estimator and Fuzzy Optimization Technique for Sensorless Induction Motor Drives

The speed estimation technique of induction machines has become a non-trivial task. For estimating the speed of an induction motor precisely and accurately an optimum state estimator is necessary. This paper deals with the performance analysis of induction motor drives using a recursive, optimum state estimator. This technique uses a full order state space Extended Kalman Filter (EKF) model where the rotor flux, rotor speed and stator currents are estimated. A major challenge with induction motor occurs at very low and at near zero speed. In such cases, information about the rotor parameters with respect to stator side become unobservable while using the synchronously rotating reference frame. To overcome this lost coupling effect, EKF observer linearizes the non-linear parameter in every sampling period and estimates the states and machine parameters simultaneously. The proposed algorithm is tuned to obtain least error in estimated speed. Any error found is further optimized using a non-linear fuzzy controller to obtain improved performance of the drive