Abstract: The accuracy of battery models is of vital importance for the health management and fault diagnosis of the battery system.With the development of battery technology, electrochemical models have become a research hotspot in recent years. The electrochemical model consists of several partial differential equations, so the model complexity is high. To reduce the complexity,the reformulated electrochemical model and the simplified electrochemical model are proposed. A comparative study of different electrochemical models can provide guidance for real-world applications. However, the thermal effects on battery electrochemical behavior are ignored in most current studies, and these studies were often conducted based on simulation. To solve the above research gaps, a comparative study of different electrochemical models is proposed based on theoretical analysis, numerical simulation, and experimental validation. Using simulation, the error source of different models is found. Using parameter sensitivity analysis, both the accuracy and efficiency of parameter identification are improved. Through electrochemical-thermal coupled modeling, the thermal effects are considered, and experiments ranging from-10 ℃ to 45 ℃ are carried out to validate model accuracy.Resultsshow that the uniform distributed reaction current assumption is the main error source for the simplified electrochemical model. This assumption works well under low current rate conditions but will cause large errors under high current conditions. Besides, incorporating thermal models can increase the accuracy of electrochemical models under different ambient temperatures. Based on experiment data, the non-simplified electrochemical-thermal model can maintain high accuracy under different ambient temperatures and working conditions, the root mean square errors of voltage were less than 25 mV. The simplified electrochemical-thermal model has high accuracy under low current rate conditions, but its accuracy decreases significantly under low state-of-charge and high current rate conditions, and the maximum root mean square error of voltage exceeds 50 mV.

Key words: battery management system, lithium-ion battery, electrochemical-thermal model, parameter identification