Position Offset Injection Based Winding and Permanent Magnet Temperature Decoupled Estimation for PMSMs

This paper proposes a decoupled approach for winding and permanent magnet (PM) temperature estimation of SPMSMs based on high frequency (HF) position offset injection. The decoupled temperature estimation model in the position-offset frame is derived, which can achieve decoupled estimation of the winding and PM temperatures. To avoid the parameter dependence and fundamental components influences, HF position offset is injected into machine to induce HF response, from which temperature dependent parameters can be extracted. In the developed estimation model, the temperature dependent terms contain cosine operation of current angle, which can ensure the high signal to noise ratio, especially in low current angle conditions. Moreover, the winding temperature estimation is decoupled from the PM temperature estimation, which can improve the accuracy of temperature estimation. The proposed approach is validated on the simulations and experiments.

I. Introduction

Accurate temperature is critical to achieve high-performance and reliable control for surface permanent magnet synchronous machines (SPMSMs) [1]–[3]. On one hand, excessive temperature rise of permanent magnetic (PM) can cause the permanent demagnetization of PMs and thus break the motor [4], [5]. On the other hand, the stator winding (SW) temperature is crucial information to prevent winding insulation breakdown [6], [7]. In other words, both the SW and PM temperatures are important for accurate and reliable control of SPMSM. As direct measurement of PM temperature is not feasible due to installation cost, accuracy and reliability concerns. Thus, PM temperature estimation has been widely investigated. Besides, estimation methods for the stator winding temperature have also been widely developed to remove the stator winding temperature sensor or improve the sensor reliability [8]–[10]. Generally, there are mainly two ways to estimate SW and PM temperatures, which are the invasive and noninvasive approaches respectively.