Multi-Objective Approach for Efficient Grid Resources Allocation in Electric Vehicle Charging Schedules
Conférence : Communications avec actes dans un congrès national
Our study introduces a novel multi-objective optimization model for electric vehicle (EV) charging scheduling, balancing two critical objectives: maximizing the energy delivered to clients while minimizing peak energy consumption. This is achieved under real-world constraints, including limited charging infrastructure, varying charging power levels, client availability, and the sequential nature of vehicle charging. The proposed approach is directly applicable to optimizing EV charging in urban environments, helping to alleviate grid stress while improving user satisfaction.
Given the NP-hardness of the problem, we employed a metaheuristic approach to approximate the Pareto front and identify optimal trade-offs. Specifically, we developed a multi-objective version of Cuckoo Search with Genetically Replaced Nests (Cuckoo-GRN), which we refer to as MOCS-GRN, and compared it to the standard multi-objective Cuckoo Search (MOCS). Both methods were adapted to our problem through carefully designed operators.
Experimental results show that MOCS-GRN consistently outperforms MOCS across three key metrics: hypervolume coverage percentage, dominance percentage, and execution time, with small standard deviations ensuring robustness. These findings demonstrate that our approach provides an effective solution for optimizing EV charging station resources, balancing grid stability and user needs.