Efficient State of Health Estimation of Lithium-ion Batteries Using Differential Thermal Voltammetry Analysis

An innovative approach for evaluating the state of health (SOH) of lithium-ion batteries (LIBs) is presented by integrating linear regression, density-based spatial clustering of applications with noise (DBSCAN), and differential thermal voltage (DTV) analysis. DTV analysis provides extensive insights into the thermal and electrochemical behavior of LIBs during charge and discharge cycles, facilitating the monitoring of degradation processes while DBSCAN effectively finds high-density regions in DTV data, revealing significant electrochemical changes within the battery. The proposed approach is emphasized for its efficiency and simplicity, making it suitable for applications in Battery Management Systems (BMS) with restricted hardware. Significant benefits are offered by this method compared to machine learning techniques such as Long Short-Term Memory (LSTM) and Gaussian Process Regression (GPR), including reduced computational and memory requirements, which are critical for real-time applications in constrained environments. Employing the experimental data, Mean Squared Error (MSE) consistently below 0.065% was achieved. The results demonstrate how well this novel approach maps a state of health (SOH) to a health index (HI), with low computational overhead and promising accuracy.