BATTERYLIFENET: A MACHINE LEARNING APPROACH FOR ACCURATE PREDICTION OF ELECTRIC VEHICLE BATTERY LONGEVITY

Abstract

The rapid growth of electric vehicle (EV) technology has intensified the demand for reliable and intelligent methods to predict battery health and lifespan. Lithium-ion batteries, which power most modern EVs, experience gradual degradation over time due to electrochemical wear, environmental stress, and operational variability. Accurate prediction of the Remaining Useful Life (RUL) of these batteries is critical for optimizing performance, ensuring safety, and minimizing maintenance costs. This research introduces BatteryLifeNet, a machine learning–based framework designed to forecast the longevity of EV batteries using real-time operational data and historical degradation patterns. The proposed framework employs a hybrid modeling approach that integrates feature engineering, health index estimation, and timeseries prediction using advanced algorithms such as Random Forest, Gradient Boosting, and Long Short-Term Memory (LSTM) networks. BatteryLifeNet analyzes voltage, current, temperature, and charge-discharge cycle data to capture degradation trends and predict future capacity fade with high precision. The model is trained and validated on standardized EV battery datasets, ensuring robustness and generalization across different battery chemistries and operating conditions. Experimental results reveal that BatteryLifeNet achieves superior prediction accuracy compared to conventional regressionbased approaches, significantly reducing estimation error and improving reliability in lifecycle forecasting. The framework not only supports efficient battery management and replacement scheduling but also contributes to sustainable energy usage by extending battery service life