Online Permanent Magnet Temperature Estimation in Permanent Magnet Machines Using Nonlinear Observer With Enhanced Parameter Estimation to Minimise Risk of Demagnetisation

Abstract

Real-time temperature estimation is critical in the control of traction drives for electric vehicles. Efficient control of conventional and novel PMSM machines requires awareness of permanent magnet (PM) temperature to avoid demagnetisation of the magnetic material. This article proposes an online parameter estimation scheme that utilises a nonlinear RISE observer to indirectly estimate the PM temperature of PMSM machines with fast convergence and high accuracy while requiring no a priori knowledge of motor parameters, no additional sensors, and no modifications to an existing motor design. In this work, nonrare earth materials are considered. This particular non-RE material is more sensitive to demagnetisation than RE material. The system architecture and analysis are presented in this article. The scheme is validated through experiments where FEA results and a rapid control prototyping (RCP) test bed are used to demonstrate proof of concept.