LPTN-Based Real-Time Stator Hotspot Temperature Estimation for Enhanced Thermal Management in High-Performance PMSMs

Thermal management is critical for high-performance electric motor drives with compact design and extreme overload capability, resulting in significant temperature gradients within the stator winding. Due to limited access to the winding hotspot, conventional temperature sensors cannot directly measure the peak hotspot temperature, which can fluctuate rapidly during operation and often exceeds accessible measurements. This constraint requires high thermal safety margins, limiting, in practice, the motor’s peak torque potential. This work deals with hotspot temperature monitoring and prediction for synchronous motor drives. This paper presents a real-time hotspot temperature monitoring solution using an advanced Lumped Parameter Thermal Network (LPTN) model. The proposed LPTN is analytically solvable, enabling efficient implementation on automotive-grade microcontrollers. Parameters are calibrated through simple experimental tests, eliminating the need for detailed motor geometry knowledge. Experimental validation on a high-performance traction PMSM under real driving conditions shows that the model achieves a residual estimation error of approximately 5° C, demonstrating reliable tracking of the winding hotspot during severe thermal transients.