A Novel Hybrid Analysis Method of LLC Resonant Converter for High Accuracy and Low Computational Burden

Abstract

Recently, as the demand for LLC resonant converters has increased in various power electronics industry sectors, research on accurate analysis methods necessary for effectively designing converters has been conducted. This paper proposes an analysis model that can estimate the efficiency of the converter and the temperature of component parts with high accuracy while requiring a relatively low computational burden when analyzing LLC resonant converters. The proposed analysis model performs circuit and thermal analysis using advanced mathematical modeling methods based on time analysis, and uses a hybrid structure analysis model that analyzes magnetic components through an AI-based analysis model. Time analysis-based analysis is a method that accurately estimates the magnitude of voltage and current of resonant converters by equivalent modeling of circuit configurations for each operating state during a switching cycle using nonlinear circuit equations. The AI-based analysis model is a method that generates the relationship between design input variables and output variables of magnetic components through AI algorithms and uses this as an analysis model. To verify the effectiveness of the proposed method, a 1kW LLC resonant converter prototype is used, and the efficiency and temperature estimation accuracy of the proposed method are measured under various capacity conditions, input and output voltage conditions, resonant network conditions, and load conditions. The proposed method estimates efficiency with an error within 2% and estimates temperature with an error within 10% under various verification conditions. In terms of computation time, it takes about 7000[s] for 100 analyses, which shows relatively low computational burden compared to existing methods. Through this, the proposed method can effectively analyze LLC resonant converters with high accuracy and relatively low computational burden.