Synergizing Knowledge and Data for AI-assisted Condition and Health monitoring in Power Electronics-Slides

Abstract: Artificial intelligence (AI) has been investigated in the power electronics sector since the 1990s, primarily for design, intelligent control, predictive maintenance, etc. Nevertheless, the initial promise of competitive AI-driven solutions has not materialized as expected, and their adoption within the industrial sector has remained limited. Presently, this landscape has undergone a substantial transformation along with cutting-edge AI tools that can synergize well-established knowledge and data. This shift is particularly relevant to condition and health monitoring applications as PES increasingly evolve into data-intensive systems with various IoT devices. The primary objective of this webinar is to offer a systematic overview of AI-assisted condition and health monitoring applications for power electronics. Situated at the intersection of data science and power electronics, it will start with an introduction to AI-assisted data-driven applications for PES. The subsequent part will present several representative case studies, including structured data-driven approach for temperature estimation, physics-informed machine learning and digital twin applications for condition monitoring, and information-fusion-based method for remaining useful life prediction. The algorithm implementation on smart hardware implementation (e.g., STM32 MCU, FPGA) will be demonstrated as well. It concludes with a discussion of ongoing initiatives and prospects for novel AI tools and open-source datasets within this dynamic and synergistic domain.