Implementing Module Health Monitoring in EV Traction Inverters

Abstract:
The traction inverter in an electric vehicle (EV) is a critical component and its unexpected failure during operation can result in a dangerous situation for the occupants. This paper describes a novel approach to add accurate Lifetime Prediction Functionality (LPF) to traction drives based on actively monitoring various device parameters through on-chip sensors. An enhanced method for predicting the remaining useful life (RUL) of a system involves monitoring the percentage shift of various health parameters, such as R(ds-on), V(th), and T(j), and analyzing them using principal component analysis (PCA). This approach improves the accuracy and reliability of the lifetime prediction model, building upon the traditional Coffin-Manson cycle-based calculation. It is important to note that relying on a single method for calculating lifetime usage is inadequate. To ensure effective online power switch health monitoring in the future, a combination of multiple methods is necessary. The LPF requires the semiconductor manufacturer to construct a database of lifetime parameters. Data is collected from various stages of the qualification process, including power module qualification, system level qualification, motor emulator testing, and the final phase where the LPF is tested and finetuned for a specific powertrain application.