A Joint Estimation Method for Lithium-ion Battery State of Charge and Capacity Considering Model Parameter Sensitivity

Accurately estimating the state of charge (SOC) and capacity of lithium-ion batteries under varying ambient temperatures and aging conditions is essential for effective battery management system (BMS). However, battery characteristics change with temperature and aging, and updating model parameters in real time will impose a computational burden on BMS. Therefore, a joint estimation method for battery SOC and capacity is proposed based on the sensitivity of model parameters to SOC estimation error. First, an equivalent circuit model is developed to represent battery's dynamics, and model parameter sensitivity analysis is performed by comparing the effect of parameter bias on SOC estimation accuracy. Second, a hybrid framework integrating long short term memory networks with Peukert equation is proposed for battery capacity estimation across various temperatures and discharge profiles. Finally, the adaptive unscented Kalman filtering based SOC estimator is designed, and the dynamic profile experiments with various temperatures and aging conditions are adopted to comprehensively evaluate the performance of proposed method. And the impact of temperature-dependent capacity correction on the SOC accuracy is analyzed in terms of estimation errors, which indicate the superiority of the method with temperature correct, especially at low temperatures and aging conditions.