Artificial Intelligence (AI) has become a promising force driving advancements in power electronics (PE) design, which covers but not limits to the optimization of discrete devices like capacitors and magnetic components, power semiconductors, hardware prototype, EMI filter, thermal management, modulation and control, etc. This seminar explores AI applications’ evolution in the PE design, including foundations, latest research progress, and future directions, with an emphasis on discussing cutting-edge advancements in depth. Beginning with an introduction to AI applications in PE including basic concepts, the session will progress to more focused discussions on AI-empowered PE design. Key challenges in existing AI algorithms, such as data scarcity, model explainability, and application transferability, will be elaborated. Afterwards, the seminar will delve into the latest innovations, physics-in-architecture neural network (PANN), which is a lightweight, explainable, and flexible AI tailored for PE design with live code demonstrations to deploy PANN for Dual-Active-Bridge converter design. Furthermore, PE-GPT, a revolutionary design tool that leverages PE-specific generative AI agents for linguistically-guided autonomous design will be presented with interactive on-site code experiments. Finally, the session will discuss forward-looking future research, with autonomous and ethical AI systems for a new PE design paradigm. This seminar is tailored for diverse groups of audiences, offering comprehensive overview about AI in PE and its design for newcomers, advanced methods and concepts for experts, and hands-on interactive code sessions to drive innovation in personal projects. This seminar aims to inspire more researchers and industry professionals to leverage AI to further drive the advacements of the industry.
