Abstract: In order to solve the large-scale real-time optimization scheduling problem of electric vehicles, a new method is proposed for dividing electric vehicles into several state clusters based on establishment of the electric vehicle state matrix. The power satisfaction and time satisfaction are defined in this paper, and their weighted sum is the user’s satisfaction. By taking the maximum economic benefit and satisfaction as the goal, a “wind-network-vehicle” real-time joint accommodation scheduling model is constructed based on the electric vehicle clustering. Aiming at the problem of grid-connected wind farm output and the uncertainty of electric vehicle loads, this paper studies the fusion of credibility theory and fuzzy opportunity constraints, introduces credibility measures, and makes clear and equivalent treatment of fuzzy opportunity constraints. Finally, a particle swarm optimization algorithm with a shrinkage factor considering the outlier penalty function method is used to optimize the scheduling model, and a case is used to verify the superiority of the model and its scheduling strategy.

Key words: electric vehicle clustering, “wind-network-vehicle” joint accommodation scheduling model, credibility theory, particle swarm joint accomodation algorithm, outlier penalty function method