Development of a Pareto-Frontier-Based Optimization Framework for Component Selection and Design Trade-Off Analysis in Power Converter Systems

Power converter design requires simultaneous consideration of efficiency, power density, thermal limits, and cost under practical hardware constraints. This paper presents a Pareto frontier based mixed integer optimization framework for component level power converter design and trade off analysis. The proposed method integrates discrete commercial component selection with multi domain analytical modeling of electrical loss, thermal behavior, converter volume, and economic cost within a unified multi objective formulation. Continuous operating parameters and discrete device choices are optimized concurrently using a modified NSGA-II algorithm supported by a surrogate neural network model to reduce computational effort. Each candidate solution corresponds directly to an implementable hardware configuration derived from a structured component database. A 1 kW DC-DC converter case study demonstrates the generation of a non dominated Pareto set representing trade offs among efficiency, power density, cost, and thermal stress. Sensitivity evaluation and ranking procedures further translate the Pareto frontier into structured design recommendations. The results confirm the capability of the framework to support systematic, implementation oriented component selection and multi objective tradeoff analysis in power converter systems.