Square Root Unscented Kalman Filters for State Estimation of Induction Motor Drives

This paper investigates the application, design, and implementation of the square root unscented Kalman filter (UKF) (SRUKF) for induction motor (IM) sensorless drives. The UKF uses nonlinear unscented transforms (UTs) in the prediction step in order to preserve the stochastic characteristics of a nonlinear system. The advantage of using the UT is its ability to capture the nonlinear behavior of the system, unlike the extended Kalman filter (EKF) that uses linearized models. The SRUKF implements the UKF using square root filtering to reduce computational errors. We discuss the theoretical aspects and implementation details of the SRUKF for IM drives. Experimental results for a direct-torque-controlled drive are presented for a wide speed range of operation, with focus on low-speed performance. A comparison with the conventional EKF and the UKF is included. Our results show that the SRUKF is a viable and powerful tool for IM state estimation.