Système Intelligent pour l'optimisation de la Répartition d'Energie dans les véhicules électriques.

Intelligent energy management is a key challenge in optimizing the range and sustainability of Battery Electric Vehicles (BEVs). Given the variability of driving conditions and the diversity of energy demands (propulsion, thermal comfort, auxiliary electronics, electronic control units, lighting systems, infotainment systems, sensors, and onboard connectivity), it is essential to design supervisory systems capable of dynamically allocating available energy according to the vehicle's dynamic priorities, environmental constraints, and driver preferences to ensure optimal performance across various driving contexts.In this context, the present thesis proposes an intelligent energy supervision architecture for BEVs, designed to optimize energy allocation between key consumers (notably the traction motor and the air conditioning system) based on operational constraints and driving conditions. Built on a multi-agent modeling framework, the architecture combines fuzzy logic and reinforcement learning to enable cooperative, adaptive, and context-aware management. The proposed controllers were evaluated using real-world driving scenarios, demonstrating a significant improvement in energy efficiency, particularly under critical conditions such as low ambient temperatures or highly variable urban routes.This work paves the way for future deployment of intelligent energy supervisors in BEVs, promoting cooperative management between agents, integration of new driving data, and anticipation mechanisms to enhance overall performance.