A comparative study of parameter identification methods for equivalent circuit models for lithium-ion batteries and their application to state of health estimation

Accurate estimation of the battery state is a crucial requirement for advanced battery management systems (BMS). Model-based state estimation methods represent the most promising option to meet BMS requirements, where the equivalent circuit model (ECM) is an effective balance between modelling complexity and accuracy. ECM's accuracy is influenced by the combination of chosen model type and parameter identification method. In this paper, batteries are aged under various conditions. Both frequency and time domain measurements are performed on batteries in a variety of aging states. These measurements are employed for comparing all combinations of 7 existing models with 7 common identification methods. In addition, the accuracy of SOH models based on ECM parameters is investigated. The experimental results indicate that for frequency and time domain measurements, the same identification algorithm may exhibit distinct performances. Overall, PSO, GWO and LSQ are ideal candidates. Among them, PSO and GWO perform optimally in the frequency domain environment, while LSQ is superior in the time domain environment. Furthermore, this conclusion does not change with battery aging. Meanwhile, a simpler model structure is even beneficial for efficiently monitoring SOH when utilizing the aforementioned superior identification methods. • Provides insights into multiple combinations of models and identification methods. • Determination of ideal identification methods with numerous battery lifespan tests. • Evaluate methods' performance in both the frequency and time domains. • ECM parameters' performance for SOH estimation are analyzed.