Adaptive model parameter identification for lithium-ion batteries based on improved coupling hybrid adaptive particle swarm optimization- simulated annealing method

The precise and robust parameterization of the battery models are of crucial important to improve safety and efficiency of electric vehicles and other applications. However, the traditional parameter identification (PI) methods usually suffer from the inaccuracy and poor robustness due to their limited searching solution. In this article, a coupled hybrid adaptive particle swarm optimization-hybrid simulated annealing (HA-PSO) algorithm along with diverse improvements is promoted for precise and robust PI process. Three categories of equivalent circuit models are performed to validate the precision and adaptability for PI on three different types of batteries, and the simulation results confirm an excellent consistency with experimental data which can satisfy the requirement of battery management system (BMS). Additionally, the numerical analysis demonstrates that the method has a satisfactory convergence speed and reasonable distribution based on Monte Carlo method. These results confirm that the presented method can be used as an effective tool for parameterizing the battery model, delivering great potential to predict battery states and other related functions based on digital technologies and cloud-control platform.