Optimizing Electric Vehicle Charging Infrastructure: A GNN-TSP Approach

Abstract

This study aims to improve the transportation sector by leveraging Graph Neural Networks (GNN) and solutions to the Traveling Salesman Problem (TSP) to enhance the deployment of charging stations in an urban environment. We focus on the city of Bucharest where we use a dataset with 153 existing charging stations and 220 potential locations for new charging stations and make use of GNN to rank the latter based on suitability. The proposed model takes into account geographical and infrastructural data and predicts the new charging stations in a non-conventional manner. Subsequently, we apply various TSP solvers to find the optimal sequence for installing the new stations, ensuring spatial efficiency. This research aims to set a new benchmark for electric vehicles charging stations infrastructure, as well as showcase the importance of AI in smart city planning. Our work can offer, in the same time, guidance for urban planners and stakeholder in the EV ecosystem.

Authors

* External Author