Multi-objective optimization for smaller, efficient and better performed design of buck-boost converters

Abstract

Converters are essential components in DC-DC transformation and each objective of a converter plays an important role in the transformation. However, in most cases, improving one objective means sacrifices the others. As a result, the overall performance of the converters is not satisfied. The thesis proposes a method to optimized volume, efficiency, and cut-off frequency of LC-filter in buck-boost converter with full consideration of keeping three objectives on optimal conditions compared to the existed method. The Multi-objective optimization is for building a more portable, highly efficient, and better performance converter. For avoiding the interference of improving each objective and for obtaining optimal solutions with a fast process and better convergence, the author applies Non-dominated Sorting Genetic Algorithm-II to generate a Pareto frontier which could provide researchers a visualized figure to select the cases based on their demands. The multi-objective optimization results are compared with single-objective optimization results to verify the feasibility of the project.