Sampled-Data Nonlinear Observer Design for Sensorless Synchronous PMSM

Abstract

The problem of designing sensorless nonlinear observers for permanent magnet synchronous motors (PMSMs) has been given a great deal of interest, especially over the few last years. In this respect, various classes of nonlinear observers have been proposed, but all of them require continuous-time measurements of the machine outputs and most of them are either designed on the basis of simplifed approximate model (not accounting for all machine nonlinear dynamics) or are not backed by a technically sound convergence analysis. Designing sampled-data sensorless observers, based on the machine nonlinear model and backed by a formal convergence analysis, has yet to be solved. In this paper, a solution to this problem is developed by combining ideas from the high gain design approach and and an inter-sample time predictor. Compared with classical high-gain observers, the new sampled-data sensorless observer includes an inter-sampled time predictor to allow large sampling period and play a crucial role in improving and enhancing the observer performances.