High-performance feedforward control of IM using speed and disturbance torque observer with noise reduction of shaft encoder

In this paper, the performance of a new speed and disturbance torque observer applied to the closed loop control of an induction machine is studied. The main novelty is that the analysis of the nonlinearity, introduced by the optical encoder quantization, allows improvement of the behavior of the filter in all the speed range. Unlike the fixed sampling frequency filters, the state estimate observational update takes place at the quantized position signal discontinuities. In this observer, the measurement noise covariance is not constant being a function of the shaft speed. Consequently, the selection of the sampling instant results in a global reduction of the output noise covariance. The observer adapts its gain depending on the state variables. The optimal gain calculation results in an observer bandwidth increase that gives a higher speed closed loop performance. Another advantage is that the self-sustained oscillations are suppressed, obtaining asymptotic stability. This is a simple method to take advantage of all the information provided by the position transducer. The proposed implementation requires an additional counter for storing the time from the encoder output transition to the actual sampling instant with regard to the usual observer, in which one counter is needed to store the actual position.