A Family of Adaptive Position Estimators for PMSM Using the Gradient Descent Method

Abstract

This article presents an evaluation of a family of adaptive position estimators (APEs) for anisotropy-based self-sensing control (SSC) of permanent magnet synchronous machine (PMSM). The APEs use the gradient descent method (GDM) to minimize a cost function and thus estimate the rotor position. Three different APEs have been developed in previous work and are revised in this article. Their dynamic performance is experimentally evaluated and compared with each other and with an encoder-based control using the same test bench. Simplified models are derived, which allow an analytic calculation of the individual speed control loop dynamics. The speed control bandwidth using the latest APE reaches almost 50% of when using a 2000 slot incremental encoder. In addition, this version of APE with GDM features parameter tuning based on analytic equations.