Global parametric sensitivity analysis of equivalent circuit model based on Sobol' method for lithium-ion batteries in electric vehicles

It is well established that the model parameters of an equivalent circuit model are crucial to improve the accuracy and stability of state estimation for lithium-ion batteries. However, real-time updates for all parameters in a traditional equivalent circuit model over the entire life cycle of the lithium-ion batteries are computationally costly owing to the excessive calculations required. To address this issue, in this study, global parameter sensitivity analysis of a typical equivalent circuit model, namely 2RCH (second order resistance-capacitance with one-state hysteresis), is performed on two types of lithium-ion batteries using the Sobol' method to investigate the global parametric sensitivity under different aging degrees. Then, the 2RCH model is simplified. The simplified 2RCH model only requires half of the model parameters to be updated regularly compared with the original 2RCH model, while the other model parameters are fixed; this significantly reduces the extent of calculations required for state estimation. Our experimental results indicate that: (1) the first-and higher-order sensitivity indices of each parameter are appropriate; (2) the simplified 2RCH model has almost the same accuracy as the original model wherein all parameters are updated, but requires only half the calculations as that in the original model. Hence, our proposed approach is of great significance as it can reduce at least half of the calculation in the state estimation over the whole battery life cycle under dynamic conditions. Also, the proposed model can improve the global accuracy of battery state estimation, which is beneficial to improve the life of battery and electric vehicle. (c) 2021 Elsevier Ltd. All rights reserved.

Automated summary

This study investigates and simplifies the equivalent circuit model of two types of lithium-ion batteries using global parameter sensitivity analysis, showing that the simplified model has similar accuracy to the original model but requires only half the calculations.