Skip to main content
APEC 2025 Main banner
Back

Technical Dialogue

Power Silicon MOSFETs, BJTs, IGBTs, etc.

D07.9 - Physics-Informed Neural Network Approach for Early Degradation Trajectory Prediction of Power Semiconductor Modules

Thursday, March 20, 2025
11:30 AM – 1:30 PM ET
Location: Level One, Hall A3
DC power cycling tests in semiconductor modules induces repetitive thermal-mechanical stresses and accumulate as fatigue over time. This digest proposes a physics-informed neural network method to reduce the reliability testing time for power semiconductor modules.The main objective of this study is to improve the accuracy of lifetime predictions using physics-informed data-driven methods, while reducing the impact of test noise and inconsistencies from device to device, and reducing the dependence on large dataset. A total of 18 IGBT devices were tested in the power cycling experiments. The proposed lifetime model can achieve an average prediction accuracy of 10% using approximately 40% of the early degradation testing data, which can reduce the testing time by about 60%.