This paper presents a cooperative estimation approach for SOC and SOH with the consideration of temperature and aging. The co-estimation is realized by using a co-estimator which is a combination of a modelbased algorithm and a data-driven technique for joint SOC and SOH estimation, and by a developed thermoelectro-aging coupling model which can reflect the dynamic and static characteristics of the parameters related to the SOC and SOH. In this co-estimator, unscented Kalman filter (UKF) and long short-term memory recurrent neural network (LSTM RNN) are designed to estimate SOC and SOH respectively and update mutually as temperature and current input; an optimized dual-time scale strategy based on slowvarying SOH and fast-varying SOC characteristics is implemented. Simulation results indicate that compared with the popular dual EKF and the UKF, the proposed algorithm gains higher accuracy and faster error convergence speed, and its estimate error for SOC and SOH are statistically less than 0.4% and 0.21% respectively under a wide range of condition.
Keywords lithium-ion battery, battery model, state of charge, state of health