关键词: 集总参数热网络, 永磁同步电机, 无迹卡尔曼, 常微分方程, 温度估算

Abstract: To address the issue of inaccurate rotor temperature estimation in oil-cooled permanent magnet synchronous motors under varying operating conditions,this paper proposes an optimized method based on a lumped parameter thermal network and an unscented Kalman filter. The study establishes a six-node lumped parameter thermal network model incorporating key components and innovatively identifies nonlinear parameters,such as thermal resistance and capacitance, using ordinary differential equations. The unscented Kalman filter is utilized to process the high-fidelity nonlinear system for real-time rotor temperature estimation. Experimental results show that even under dynamic conditions with sudden load changes, the estimation error of the proposed method remains around 1℃ , demonstrating high accuracy and strong robustness. This approach offers an effective sensorless monitoring solution for the precise thermal management and safe operation of high-power-density oil-cooled motors.

Key words: lumped parameter thermal network, permanent magnet synchronous motor, unscented Kalman filter, ordinary differential equations;temperature estimation