Battery SOC and SOH Estimation Based on Sage-Husa Extended Kalman Filter and Feedforward Neural Network

This paper proposes a unified framework for accurate estimation of battery State of Charge (SOC) and State of Health (SOH) that integrates the Sage-Husa Extended Kalman Filter (EKF) with a Feedforward Neural Network (FNN). Adaptively estimated process and measurement noise covariances from the Sage-Husa EKF are employed as dynamic input features for the FNN, enhancing robustness to varying noise conditions. The framework is validated using 300 cycle lithium-ion battery data in both Model-In-the Loop Simulation (MILS) and Software-In-the-Loop Simulation (SILS) environments in MATLAB/Simulink. Results demonstrate that the Sage-Husa EKF-FNN achieves outstanding SOH prediction accuracy, with a Root Mean Squared Error (RMSE) of 0.0030676% and a Mean Absolute Error (MAE) of 0.0009185%, surpassing conventional methods. In SOC estimation, the Sage-Husa EKF attains the highest accuracy, with an RMSE of 0.0279150% and an MAE of 0.0011811%. The consistent MILS and SILS results confirm the reliability of the proposed approach for embedded applications, establishing a solid foundation for future real-time validation in advanced battery management system.