Nonlinear speed observer for high-performance induction motor control

In this paper, we consider the problem of estimating the angular velocity of an induction motor using encoder measurements. Two methods are compared. In the first method, the speed is found by calculating the backward difference of the position measurement and low-pass filtering the result. In the second method, the velocity is estimated using a nonlinear observer constructed using the known dynamic model of the induction motor. The performance of the two methods is evaluated in the context of their use for velocity feedback in a high-performance field-oriented control law. Experimental results demonstrate that the speed observer leads to a smoother operation of the motor in closed-loop. With the estimator based on differentiation, either the delay imposed by the low-pass filter is too large to maintain high bandwidth feedback, or the fluctuations in the estimated speed are so large that much more energy ends up being dissipated to achieve the same control task.