Efficient Location of Electric Vehicle Charging Stations Using Traffic Flow Simulations and Artificial Intelligence

This study addresses the optimization of Electric Vehicle (EV) charging station placement in urban settings, balancing travel time, number of active stations, and quality of service. Using a multi-objective evolutionary algorithm integrated with the MATSim simulation toolkit, we modeled real-world conditions in the city of Cuenca, Ecuador, generating a dataset of 1,200 unique entries with 20 predictor variables and 3 target variables that reflect diverse configurations and operational demands. This data served to train various regression models, linear methods as well as more robust approaches like Random Forest and Neural Network, to evaluate their effectiveness in capturing EV infrastructure requirements. Results indicate that Neural Networks outperformed other models with R<sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup> values exceeding 90% for some variables, while Random Forest also performed well. Linear models, however, struggled with non-linear relationships, especially in predicting quality of service, where all models showed higher MSE values, underscoring the complexity of this variable. Despite these challenges, the dataset supports advanced modeling techniques, providing a valuable foundation for strategic EV infrastructure planning.