A Digital Twin Framework for Dual-Path Estimation of Battery State of Charge and State of Health Using Hybrid Physics-Informed Machine Learning

Digital Twin (DT) technology enhances conventional Battery Management Systems (BMS) by creating dynamic virtual models that replicate physical assets. This paper presents a novel synchronized dual-path DT framework for integrated estimation of State of Charge (SOC) and State of Health (SOH) in lithium-ion batteries, addressing the limitation of existing methods that treat SOC and SOH separately. The proposed dual-path architecture enables their collaborative integration within a synchronized DT. A hybrid Kalman Filter-Long Short-Term Memory (KF-LSTM) model is employed for robust SOC estimation, while Random Forest Regression is applied for SOH prediction. This physics-informed machine learning approach ensures physically consistent short-term state tracking while capturing nonlinear degradation patterns for accurate long-term health prediction. The framework was validated using 600 experimental charge-discharge cycles. Results show that the hybrid SOC estimator achieved a Root Mean Square Error (RMSE) of 2.4%, and the SOH predictor attained an R² score of 0.98. The results show the proposed DT can reliably track battery states in real-time while supporting predictive health management. This makes it a practical basis for future storage solutions in electric vehicles.