This study presents a novel methodology for state-of-charge (SOC) prediction, integrating kinematic and environmental data with Vissim and SUMO traffic microsimulation platforms. The approach exploits the complementary advantages of both tools—Vissim’s behavioral realism and SUMO’s computational efficiency—to build a flexible framework applicable in diverse urban environments. The machine learning model, based on the XGBoost algorithm, achieved high predictive accuracy (R² = 0.86; RMSE = 7.213) using only four input variables: vehicle speed, acceleration, road gradient, and ambient temperature. This minimalist design removes the dependency on expensive sensors on board while preserving robustness in large-scale urban simulations. Practical applications include the identification of SOC-based energy hotspots for traffic control, strategic planning of charging infrastructure, and generation of predictive energy consumption maps for urban mobility planning. The modular structure of the model allows future extensions that incorporate factors such as battery aging, humidity, or driving behavior, and supports potential real-time onboard deployment. The results contribute to the advancement of energy-conscious traffic modeling, intelligent infrastructure development, and electric vehicle fleet optimization—key pillars of sustainable urban mobility systems.