Transient Temperature Evaluation of PMSM by the Combination of LPTN and Segmented Iron Losses Model With Compensation Terms

• Published in: IEEE transactions on transportation electrification Vol. 11; no. 4; pp. 10113 - 10124
• Main Authors: Li, Jiahao, Zhao, Shiwei, Yang, Xiangyu, Cao, Jianghua, Gao, Mengzhen
• Format: Journal Article
• Language: English
• Published: IEEE 01.08.2025
• Subjects: Accuracy; Computational modeling; Frequency conversion; Iron; Iron losses calculation; lumped parameter thermal network (LPTN); Magnetic hysteresis; Mathematical models; permanent magnet synchronous machine (PMSM); Skin effect; Temperature; Transient analysis; transient temperature; Windings
• ISSN: 2332-7782; 2332-7782
• DOI: 10.1109/TTE.2025.3564495

This article presents an improved transient temperature evaluation method for permanent magnet synchronous machine (PMSM) based on the lumped parameter thermal network (LPTN) and the segmented iron loss calculation model with compensation terms. The modeling methodology for the proposed iron loss calculation model (PILCM) is given in detail, including the process of model establishment and parameter determination. By introducing the compensation term of eddy current loss and hysteresis loss in different frequencies and magnetic density ranges, the loss characteristics of the proposed model demonstrate higher computational accuracy than conventional iron loss calculation model (CILCM), which can be confirmed by magnetic characteristic testing experiments of lamination steel. In addition, an experimental platform of the temperature rise is built for evaluating the temperature of end winding, the absolute error of the LPTN + PILCM and the LPTN + CILCM under rated condition are <inline-formula> <tex-math notation="LaTeX">1~^{\circ } </tex-math></inline-formula>C and <inline-formula> <tex-math notation="LaTeX">2.3~^{\circ } </tex-math></inline-formula>C, respectively. The error of overload running time predicted by LPTN + PILCM under two overload conditions are 3.6% and 5.3%, respectively, which has higher accuracy than the results of 11.7% and 13.1% predicted by LPTN + CILCM.