Power lithium battery model parameter identification based on improved cat swarm algorithm

In order to maximize the life span and ensure the safety of lithium batteries, improving the accuracy of battery model parameter identification is crucial. Therefore, based on the traditional second-order RC equivalent circuit model, an improved strategy of the cat swarm optimization (CSO) algorithm is explored and applied to the parameter identification of lithium batteries. Initially, the time-domain response expression of the equivalent circuit model is obtained. The initial diversity of the cat swarm is enhanced by introducing the Logistic chaos mapping into the CSO, while the search range is expanded through Lévy flight, and an adaptive inertia weight is adopted to improve global search capabilities. Subsequently, battery pulse charging and discharging experiments are conducted to obtain the fitting coefficients of the OCV - SOC curve. Combining these data, the improved CSO is utilized for parameter identification. Finally, MATLAB/Simulink is used to construct the battery equivalent circuit model for simulation of the identified model parameters, and a sensitivity analysis of these parameters is performed. Experimental results demonstrate that the simulation with the improved CSO algorithm achieves a maximum absolute error reduction of 25.9% and a root mean square error reduction of 40.4% compared to the particle swarm optimization (PSO) algorithm. The improved CSO algorithm effectively enhances the accuracy of parameter identification and the precision of the battery model, showing great practicality.