Design and Implementation of Velocity Estimators for Motor Velocity Control

This paper studies the problems of design and implementation of velocity estimators for motor velocity control. In practice, most servomotors use the encoder to measure the position of the motor and then use the conventional differential algorithm, dividing the displacement between two sampling points by the sampling time, to obtain the velocity for feedback control. However, this way can result in serious noise amplification. In this study, velocity estimators are used to solve this problem. This paper compares three velocity estimators including PI Servo-loop velocity estimator, Levant differentiator, and Kalman filter. First, MATLAB/Simulink are used to simulate these velocity estimation algorithms. For further validation, these velocity estimation algorithms are simulated and tested with the actual motor position signals. In experiments, the estimators are implemented on a digital signal processor (TMS320F28335) from Texas Instruments. As a result, the Kalman filter outperforms the other velocity estimators in velocity control.