Optimization of the C2V2C model: prediction of the mobility patterns and charging recommendations

Details

Supervisors

Verhaeghe, Hélène

Faculty

Ecole polytechnique de Louvain

Degree label

Master [120] en sciences informatiques, à finalité spécialisée

Abstract

enIn the context of the energy transition, Renewable Energy Communities (RECs) aim to foster local energy production, sharing, and consumption. Integrating electric vehicles (EVs) into these communities opens new possibilities for flexible energy management. This thesis investigates the optimization of a Community-to-Vehicle-to-Community (C2V2C) model, where EVs serve as mobile energy power banks between RECs. This thesis addresses two main tasks: predicting mobility and charging behaviors of EV users and designing a smart charging (SC) algorithm that maximizes self-consumption within RECs. We explore multiple datasets describing REC consumption and EV usage patterns and use clustering and machine learning techniques to classify user behaviors and predict key variables such as plug-out time, next destination and energy consumption. Based on these predictions, we propose a decision algorithm to recommend personalized charging strategies, considering grid constraints and REC surplus availability. We validate our approach through simulations of EV trips and charging sessions. Then we evaluate the model's impact on energy self-consumption, discharging capacity to the RECs and system feasibility by monitoring the number of EVs that run out of battery. The results underscore the potential and limitations of data-driven smart charging strategies. They highlight the importance of accurate temporal predictions and suggest future work directions such as incorporating user-declared intentions, testing with real-world data and extending the methodology to support multi-EV simulations and economic optimization.