State of Charge Estimation for Lithium-ion Battery Using Recurrent Cerebellar Model Neural Network with Kalman Filter

The state of charge (SOC) is a crucial parameter for reflecting the battery's endurance. This study proposes the novel method of lithium-ion battery SOC estimation to ensure the working status of the energy storage system (ESS). Recurrent cerebellar model neural network (RCMNN) and Kalman filter (KF) are both applied for the SOC estimation that recurrent units can capture the dynamic features. The inputs of RCMNN and KF include voltage, current, and temperature for simulating the general situation of ESS. The battery data are collected in the Fujian Special Equipment Inspection and Research Institute. The results show that the accuracy and robustness of the proposed method under different conditions.