A Hybrid Approach to EV Route Optimization and Charging Demand Prediction for Sustainable Transportation

The rapid growth of urban Electric Vehicle (EV) fleets demands advanced strategies to optimize routing and charging infrastructure, as conventional methods overlook dynamic factors like battery state, real-time traffic, weather, and station capacity, resulting in wasted energy and increased emissions. We develop a 150-word hybrid framework combining a Genetic Algorithm (GA) for global route exploration with Particle Swarm Optimization (PSO) for local refinement, and employ Random Forest regression to forecast station-level charging demand using environmental, traffic, vehicle, and schedule features. On a real-world EV dataset, our GA-PSO hybrid achieves 275 kWh energy savings, reduces travel distance by 80 km and travel time by 45 minutes, and cuts average queue time by 50%, while the Random Forest predictor attains MAE = 2.99 kW and RMSE = 3.78 kW, outperforming Linear Regression. CO<inf>2</inf> emissions decrease by 192.5 kg. This integrated approach significantly enhances operational efficiency and sustainability and provides a scalable solution for future smart-mobility systems.