Globally Stable Resistance and Speed Joint Estimation with State Transformations for Sensorless IPMSM Drive

Abstract

This article addresses the nonlinear-parameterized problem of known regressor full-order interior permanent magnet synchronous motor (IPMSM) models, and proposes asymptotically stable observers for the joint estimation of stator resistance, rotor position, and speed. In recent years, several known regressor variants of nonlinear-parameterized full-order IPMSM models have been identified, which eliminate most of the low-speed instability problems. The resulting nonlinear high-gain observers, however, cannot adjust the convergence speed of parameters separately, and their parameter identifiability varies greatly with working conditions. It is, therefore, more appealing to consider a linear adaptive observer to realize stable joint estimation. In this article, by designing model reference adaptive systems and introducing the filtered regressors, the decoupling of nonlinear error dynamics is achieved. The linear feedback design makes it possible to guarantee globally asymptotic stability by applying the Kalman–Yakubovich lemma, while avoiding the disadvantages of nonlinear high-gain observers. Besides, the stability analysis results indicate that the influence of high-order errors can be eliminated by carefully tuning the observer coefficients. Several elaborate working conditions are considered to perform effective comparative experimental verification, including very-low speed experiments, wide-speed operations, and motor reversal tests.