Remaining Useful Lifetime Prediction Based on Extended Kalman Particle Filter for Power SiC MOSFETs

Area of Science:

• Materials Science
• Electrical Engineering
• Semiconductor Device Physics

Background:

• Silicon-based devices are nearing their performance limits for high-power applications.
• Silicon carbide metal-oxide-semiconductor field-effect transistors (SiC MOSFETs) are promising alternatives but face reliability challenges.
• Key reliability issues include bias temperature instability, threshold voltage drift, and reduced short-circuit robustness.

Purpose of the Study:

• To address the critical need for accurate remaining useful life (RUL) prediction in SiC MOSFETs.
• To propose and validate a novel RUL estimation method for SiC MOSFETs.
• To enhance the reliability and operational safety of high-power electronic systems utilizing SiC MOSFETs.

Main Methods:

• Development of a new power cycling test platform to monitor SiC MOSFETs.
• Utilizing on-state voltage as a precursor for degradation monitoring.
• Implementation of an Extended Kalman Particle Filter (EPF) algorithm based on an on-state voltage degradation model.

Main Results:

• The proposed EPF method significantly reduces RUL prediction error.
• RUL prediction error decreased from 20.5% (traditional Particle Filter algorithm - PF) to 11.5% with EPF, using 40% data input.
• Life prediction accuracy was improved by approximately 10%.

Conclusions:

• The EPF-based RUL estimation method offers superior accuracy for SiC MOSFETs.
• This approach enhances the reliability assessment of wide bandgap power semiconductor devices.
• The findings contribute to the development of more robust and predictable high-power electronic systems.