Impact of Loss Model Selection on Power Semiconductor Lifetime Prediction in Electric Vehicles

TL;DR

This paper compares power loss estimation models with different time resolutions for EVs, revealing that inappropriate models can lead to lifetime prediction errors up to 300 times, emphasizing the importance of model selection.

Contribution

It provides a comparative analysis of loss models with different time resolutions for EVs, highlighting the impact on lifetime prediction accuracy.

Findings

Output period average models are limited for dynamic EV driving cycles.

Incorrect loss model choice can cause lifetime prediction errors up to 300 times.

The study underscores the importance of selecting appropriate loss models for EV lifetime estimation.

Abstract

Power loss estimation is an indispensable procedure to conduct lifetime prediction for power semiconductor device. The previous studies successfully perform steady-state power loss estimation for different applications, but which may be limited for the electric vehicles (EVs) with high dynamics. Based on two EV standard driving cycle profiles, this paper gives a comparative study of power loss estimation models with two different time resolutions, i.e., the output period average and the switching period average. The correspondingly estimated power losses, thermal profiles, and lifetime clearly pointed out that the widely applied power loss model with the output period average is limited for EV applications, in particular for the highly dynamic driving cycle. The difference in the predicted lifetime can be up to 300 times due to the unreasonable choice the loss model, which calls for the industry attention on the differences of the EVs and the importance of loss model selection in lifetime prediction.