Reduced order extended Kalman filter for state estimation of brushless DC motor

Abstract

This paper presents a discrete time reduced order extended Kalman filter (EKF) for state estimation of sensorless brushless DC (BLDC) motors. Permanent magnet brushless DC motors are widely used in industries as they possess high power density and simplicity in control. The drawbacks of position sensors can be avoided by sensorless scheme. An extended Kalman filter based sensorless method is considered in this work. Computational burden of the full order Kalman filter can be minimized by a reduced order model. In this paper an extended Kalman filter algorithm is implemented using a reduced order discretized state space model which estimates the rotor speed and rotor position from the measured stator currents. The order reduction minimizes the computational time of the filter and also greatly simplifies the tuning of the covariance matrices. The reduced order filter performance is compared with that of full order filter with respect to the speed and position estimation, execution time of the filter and tuning of the algorithm.