Parameter Identification of Lithium-Ion Batteries Using an Improved Chaos Evolutionary Optimization Algorithm
Abstract
This paper proposes an improved Chaos Evolutionary Optimization (CEO) algorithm for estimating the six parameters of a fractional-order battery model integrating a constant phase element and a Warburg impedance.To enhance global search and maintain population diversity, an adaptive hybrid strategy that combines boundary shrinkage with reflection is introduced. The proposed method is rigorously validated using the Dynamic Stress Test and the Urban Dynamometer Driving Schedule across three different temperatures. Experimental results demonstrate high fitting accuracy and favorable generalization capability. Furthermore, under identical experimental settings, the improved CEO consistently achieves lower prediction errors compared with the standard Particle Swarm Optimization algorithm.The proposed approach provides a reliable high-precision solution for battery parameter identification over a wide temperature range.