Multi-Objective Design Automation in Power Electronics Using Bayesian Optimization Techniques

This paper presents a design automation framework for power converters, integrating finite element analysis (FEA) simulations with Bayesian optimization (BO) to enhance design precision and decision-making. The methodology conducts multi-objective optimization, balancing efficiency and cost, while accounting for discrete design variables such as specific semiconductor device options. Detailed FEA simulations are embedded within the optimization loop to accurately model and refine magnetic components, ensuring practical and high-performance outcomes. Engineers are provided with a Pareto front (PF) of optimized solutions, enabling an objective selection of the most suitable design. The approach is validated through a 10-kW interleaved boost converter (IBC), demonstrating its capability to deliver cost-effective, high-efficiency solutions tailored for advanced power converter converters.