Multi-Objective Parameter Identification Method for Electrochemical Model of Lithium-Ion Battery

Prediction of the battery state of health is becoming more important from the perspectives of battery design and utilization. An electrochemical model can be a feasible option to simulate the internal reaction of lithium-ions. However, it is difficult to implement a battery model because of multiple criteria to evaluate the battery voltage curve and long battery modeling time. In this paper, we propose a parameter identification method that rapidly derive the optimal parameters while satisfying multiple objectives. We suggest the cost function to evaluate multiple objectives and the search space reduction method to approach the optimal parameters. The proposed method is validated with battery charging experimental data from Samsung SDI. The validation results show that the root-mean-square error of the battery model is reduced by an average of 67.06 % than the baseline.