Modeling and Online SOC Estimation Method Using Back Propagation Neural Network with Momentum Gradient Descent Algorithm of Batteries

To improve the estimation accuracy of the state of charge (SOC) in power batteries for electric vehicles, this study proposes a novel modeling and online SOC estimation method using Back Propagation neural network with momentum gradient descent algorithm (BP-MGD). Integrating the equivalent circuit model, a BP neural network is utilized for battery modeling. SOC estimation is performed by combining Taylor expansion with BP neural network modeling. To enhance the algorithm's convergence speed, the network is trained using the momentum gradient descent algorithm. Experimental results indicate that the proposed approach achieves high accuracy, with ME, MAE, and RMSE all maintained within 0.3%.