Physics-informed Artificial Intelligence Methods for Prognostics and Health Management in Power Electronic Converters

Abstract:
High power density and high efficiency are always the design objectives of power electronic systems. Prognostics and health management (PHM) technology is a proactive measure to realize operation optimization, predictive maintenance, and high reliability of power electronic converters.
This project focuses on utilizing physics-informed AI methods to achieve condition monitoring and failure prognosis through an in-depth understanding of degradation mechanisms, physical models, and data-driven methods. Physics-informed AI approaches can seamlessly integrate data and physical knowledge and improve the generalization and interpretability of models. Utilizing historical operation data and condition monitoring data of power electronic converters adequately and combining the proposed methods to carry out state estimation, parameter identification, and health assessment is expected to be a means of discovering a new path for power electronic system reliability research and analysis.

Funding: CSC Scholarship