Project Description

This project deals with develoing a high fidelity, real-time Digital Twin that integrates physics of failure models with machine learning to predict the Remaining Useful Life (RUL) of power electronic converters and optimize their switching strategies under dynamic loading conditions. Wide-Bandgap (WBG) devices operate at high frequencies where parasitic inductances and rapid thermal cycling significantly impact reliability. There is a gap in "closed loop" twins that can not only monitor health but also adjust control parameters (e.g., switching frequency or gate drive voltage) to mitigate degradation in real time. The project would involve creating a virtual multi domain model, data synchronization and adaptation layer, and creation of reliability control algorithms. The Ph.D. work is expected to bridge the gap between microsecond switching transients and the long-term (years) aging process of power modules by combining the accuracy of physics based aging models with speed of AI based forecasting while validating the developed algorithms in real time or hardware in loop simulations.