Lithium-ion batteries state of charge estimation using PSO-BP neural network improved UKF algorithm

To tackle the issue of inaccurate state of charge (SOC) estimation in lithium-ion batteries when employing the unscented Kalman filter (UKF) in real-world applications, a refined approach utilizing the PSO-BP neural network is introduced to enhance the performance of UKF for SOC estimation. Firstly, the Thevenin model is identified using particle swarm optimization (PSO). Secondly, the PSO-BP is trained by using the characteristic parameters related to the estimation error of UKF. Finally, the UKF SOC estimate error was corrected using the trained PSO-BP. Experimental results demonstrate that the proposed algorithm significantly enhances SOC estimation accuracy.