Investigation of TSEPs based on the Feature Extraction from the Gate Current

Jiawei Liu, Hongfei Chen, Jinghan Lin, Zekun Li, Haimeng Wu, Bing Ji*

*Corresponding author for this work

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

1 Citation (Scopus)


Electric Vehicles demand high-performance, high-reliability power semiconductor modules. There is a strong correlation between junction temperature and the reliability and safety of power modules. To estimate junction temperature behaviour, a reliable, non-intrusive, and less load-affected procedure is required. This paper proposed a time-domain analysis that only focuses on the gate current channel of the SiC MOSEFT power module without the need of calibration of load current. The statistic features extracted from the turn on gate current waveform are selected and then fused by Principal Component Analysis (PCA). The KMO demonstrates the suitability of the dataset for PCA and the fused Principal Components (PCs) are fed into the Multiple Linear Regression (MLR) model to estimate temperature. The higher accuracy is generated by this statistic features fed PCA-MLR model compared with conventional gate peak current regressed model.

Original languageEnglish
Title of host publication11th International Conference on Power Electronics, Machines and Drives (PEMD 2022)
PublisherInstitution of Engineering and Technology
Number of pages5
ISBN (Electronic)9781839537042, 9781839537059, 9781839537189, 9781839537196, 9781839537615
Publication statusPublished - 29 Aug 2022
Event11th International Conference on Power Electronics, Machines and Drives, PEMD 2022 - Newcastle, Virtual, United Kingdom
Duration: 21 Jun 202223 Jun 2022


Conference11th International Conference on Power Electronics, Machines and Drives, PEMD 2022
Country/TerritoryUnited Kingdom
CityNewcastle, Virtual

Cite this