AI-Assisted Parameter Optimization Design for LLC Resonant Converter

The rapid development of transportation electrification has posed higher requirements for the power density of power electronic devices. LLC converters are widely used, and the parameters need to be carefully designed. Due to the multiple electrical and magnetic variables and their internal coupling in the converter, optimizing the total power loss is a challenge. Moreover, the voltage gain obtained by traditional fundamental harmonic analysis (FHA) becomes inaccurate when deviating from the resonant point, adversely affecting parameter design and modulation. In this work, a power loss model containing electrical and integrated magnetic component parameters is established, and a voltage gain model based on a neural network is built. The Particle Swarm Optimization (PSO) algorithm is utilized to solve the multi-variable, multi-objective optimization problem, interacting with the developed power loss model and voltage gain model to simultaneously obtain the required electrical and magnetic component parameters, aiming to optimize power loss, gain characteristics, and footprint. A 270/28V LLC converter prototype is built to verify the proposed design.