Real-time Extended Kalman and H∞ filters performance comparison for induction machine speed and torque estimation

This paper presents a performance comparison of an Extended-Kalman and Extended H_∞ filters used to estimate flux, rotor speed and torque in a PWM inverter-fed induction machine. The comparison is performed in real-time in an induction machine testbech at different speed ranges and subject to load changes. Results indicate that the H_∞ estimator overcomes the Kalman Filter in the low speed range with no significant increase in execution time using a Fixed Point DSP, while performing almost identically in the medium and high speed range.