Optimized EKF algorithm using TSO-BP neural network for lithium battery state of charge estimation

Abstract

The BP neural network can effectively improve the accuracy of state of charge (SOC) estimation by the EKF algorithm. However, the BP neural network is strongly influenced by the initial weights and thresholds, which limits its application in the SOC estimation. To improve the current defects of the BP neural network for better application to the SOC estimation, this paper proposes a method to improve the performance of the EKF algorithm for SOC estimation by optimizing the BP neural network using the tuna swarm optimization (TSO) algorithm. Based on the constructed first-order RC battery model, the optimized BP neural network was trained offline and used to compensate the SOC estimation error of the EKF algorithm online. The superiority of the proposed algorithm for SOC applications was demonstrated by simulation experiments under both dynamic stress test (DST) and Beijing dynamic stress test (BJDST) operating conditions. In addition, comparative experiments with existing hybrid algorithms were conducted. The simulation results show that compared with the EKF algorithm and the traditional BP-EKF algorithm, the proposed algorithm has better estimation accuracy and convergence for SOC and can track the SOC variation well. Compared with the conventional BP-EKF algorithm, the RMSE and MAE values of the proposed algorithm decreased respectively by 35.7 % and 37 % under DST, and respectively by 38.4 % and 43.3 % under BJDST. Besides, the proposed algorithm has better performance than other hybrid algorithms.