Digital-Twin Based SOC and SOH Estimation for Lithium-Ion Batteries

The safe and reliable operation of Electric Vehicles (EVs) depends largely on the performance and health of lithium-ion batteries. This work presents a Digital Twin framework for real-time estimation of two critical battery health parameters: State of Health (SOH) and State of Charge (SOC). A real-world dataset containing over 1,000 charge–discharge cycles is utilized, with features such as voltage, temperature, ohmic resistance (Re), charge transfer resistance (Rct), and capacity. Data preprocessing includes removal of missing values and outliers, along with normalization to enhance model accuracy. SOH is estimated based on capacity degradation and resistance growth, while SOC is derived using Coulomb counting, complemented with voltage–time analysis and Kalman filtering. These estimations are validated through a Simulink-based Electrical Equivalent Circuit Model (ECM) that replicates real-time battery voltage and current characteristics. Furthermore, multiple machine learning models including regression, exponential decay, and random forest are evaluated for SOH prediction using performance metrics such as RMSE and R². The framework is designed with modularity and scalability, enabling real-time integration into intelligent Battery Management Systems (BMS). This ensures adaptability across different EV platforms and operating conditions. The methodology also emphasizes predictive capability, supporting advanced battery diagnostics and prognostics. With the increasing demand for sustainable mobility, reliable and long-lasting energy storage is vital. However, lithium-ion batteries face progressive degradation, affecting both efficiency and safety. The proposed framework addresses this challenge by combining data-driven modeling with Digital Twin technology. Ultimately, this research contributes to the development of scalable, adaptive, and predictive battery health monitoring solutions, thereby enhancing EV safety, efficiency, and lifetime.