Minimizing EV battery degradation in V2G systems via XGBoost-guided particle swarm optimization

Electric vehicles (EVs) are increasingly integrated into smart grid infrastructures through Vehicle-to-Grid (V2G) technology, which enables bi-directional energy flow and enhances grid reliability. However, frequent charging and discharging cycles accelerate battery degradation, posing a critical barrier to widespread V2G adoption. This study proposes a hybrid optimization framework that combines the predictive capabilities of eXtreme Gradient Boosting (XGBoost) with the global search efficiency of Particle Swarm Optimization (PSO) to address this challenge. First, XGBoost models and forecasts battery degradation based on real-world EV charging data enriched with degradation indicators derived from the Arrhenius equation. These predictions are then integrated into a PSO-driven scheduling algorithm that generates optimal 24-hour charge-discharge plans to minimize battery wear while satisfying grid support requirements. Simulation results demonstrate that the proposed method limits degradation to approximately 0.0466% daily, aligning energy usage with grid demand while extending battery life. This intelligent, data-driven approach offers a practical pathway for sustainable V2G deployment—balancing grid performance, battery health, and economic viability—particularly in emerging EV markets.