Optimizing Battery Performance in Electric Vehicles Through Predictive Modeling

The fast-paced electrification of the automobile sector has intensified an immediate need for reliable battery management systems capable of sustaining long-term performance. One of the most important indicators when it comes to battery health is State of Health (SOH), which indicates how much capacity is left and how long a certain battery will endure. Reliable estimation of SOH is critical to enable the optimal performance of vehicles, reduce service costs, avoid unanticipated failures. In this project, we propose a machine learning approach to predict the SOH of EV batteries based on the two widely used algorithms, XGBoost and LGBM. In order to cover and analyze as many process variables available as possible to build a database for the training and evaluation of the models, we measured multiple battery parameters, such as voltage, current, temperature and historical charge/discharge cycles. The train test split this dataset in such a way that the greater part of the dataset is used for the machine learning operations. Individual models, such as XGBoost and LGBM, are used to be train due to its strength in handling large data and complex data with a high precision. This study attempts to find out the better of the two algorithms for real-time SOH prediction in practice by analyzing the accuracy and performance metrics of both algorithms. Results of this analysis can help manufacturers and operators make data-informed decisions regarding battery usage, maintenance schedules and replacement, which will contribute to improved vehicle reliability and extended battery life.