Multi-Objective Optimization of LLC Resonant Converters for Efficiency and Power Density Based on Time Domain Analysis and Genetic Algorithm

This article presents an optimization design methodology for LLC resonant converters, aimed at optimizing both converter losses and the volume of magnetic components. The optimization variables include transformer turns ratio, resonant inductor, magnetizing inductor, resonant capacitor, and peak flux density. In rated power conditions, the converter operates in specified operational modes, which enables soft switching and ensures monotonic voltage gain. A time-domain model of the LLC converter is discussed, and a rolling iterative method is proposed for obtaining time-domain solutions across a wide voltage gain range. Based on time-domain solutions, the loss model can be accurately calculated, and the design of magnetic components is provided. Due to the inherent tradeoff between loss and volume, a single optimal solution cannot be determined. The Non-Dominated Sorting Genetic Algorithm II (NSGA- II) is employed to optimize both loss and volume, seeking multiple Pareto-optimal solutions. Finally, experimental results demonstrate that the converter can achieve a maximum efficiency of 98% or a minimum magnetic component volume of 19.03 cm³.