Digital Twin-based and Machine Learning-assisted GaN Power Module Packaging Reliability Analysis Method

With growing global emphasis on environmental sustainability, wide bandgap semiconductors such as silicon carbide (SiC) and gallium nitride (GaN) have been widely adopted in power modules due to their superior performance over silicon devices for efficient energy conversion. Between them, GaN devices stand out in fast switching applications, driving the development of integrated packaging designs to fully exploit their potential. However, the heterogeneous structures of these designs pose significant challenges for reliability assurance, where conventional test-based approaches are limited by prototype availability and measurement difficulties inherent in compact structures.

This PhD research addresses these challenges by employing digital twin (DT) methods to gain detailed insight into the physical responses of complex packaging structures, and integrating machine learning (ML) methods to overcome computational constraints, forming a hybrid framework for reliability analysis, and offering a robust, generalizable toolset for GaN power module  reliability-oriented packaging optimization.