SOC Estimation of Li-ion Batteries Based on IHHO-KELM Model

Accurate prediction of the state of charge(SOC) of Li-ion batteries is a key condition to achieve intelligent management of battery management systems. The SOC estimation model of the improved Harris hawks optimization(IHHO) optimized kernel extreme learning machine(KELM) is proposed. Based on the standard Harris hawks optimization(HHO), a logistic chaotic mapping is introduced to obtain the optimal population of individuals and improve the algorithm’s optimization capability. The jump distance J is optimized, and the regulating operator nonlinear control mechanism is constructed to balance the exploration and exploitation behaviors, so that the proposed algorithm can be more rationalized in the pre and post search. Experimental simulations with five standard test functions demonstrate that the IHHO algorithm has better search capability, and the IHHO algorithm is used to optimize the parameters of the nuclear limit learning machine and establish the IHHO-KELM estimation model. The experimental data of constant current discharge is used for simulation study, and the prediction results of unscented Kalman filter(UKF), grey wolf optimizer-back propagation(GWO-BP) and IHHO-KELM models are compared and analyzed, and the robustness of the model is verified by choosing dynamic stress test(DST) working condition. The results show that the mean square error and mean absolute error of SOC prediction of the proposed model are reduced to 0.13% and 0.7%, respectively, and the accuracy is improved with better robustness.