An Improved Cascaded Forward Back Propagation Neural Network for State of Energy Estimation of Lithium-ion Batteries

The SOE of batteries is crucial for power system energy scheduling and new energy vehicle industry innovation, with scientific research significance. To achieve precise SOE estimation, this study introduces a data - driven high - precision SOE estimation model. First, a moving - average data enhancement technique is proposed to reduce noise and increase data diversity, improving estimation efficacy. Then, a Cascaded Forward Back Propagation Neural Network is built, and the ESC parameter optimization algorithm is used to optimize network weights and thresholds. Finally, the model is validated with CALCE battery team's public dataset. Experimental results show the model's superior performance: across different temperatures, MAE averages 0.015, MSE averages 0.0006, MAPE averages 2.29%, and R^2 averages 0.9828. The methodology's effectiveness is confirmed, providing a reliable reference for the new energy vehicle industry..