Hierarchical model predictive control-based electric vehicle fleet charging management

Due to the relatively long parking/grid-connection intervals, charging of electric vehicles is characterized by flexibility that could be exploited for different benefits such as charging cost minimization and better utilization of intermittent renewable energy sources. To this end, an optimal and predictive hierarchical charging management method of electric vehicle fleet, characterized by computational efficiency and good scalability, is proposed in the paper. The method relies on an aggregate electric vehicle fleet model, a model predictive control law that performs an on-line dynamic programming optimization of an aggregate charging power, and a heuristic algorithm that distributes the aggregate charging power to individual electric vehicles. The heuristic algorithm is set to prioritize electric vehicles with lower state-of-energy levels and sooner time-of-departure. The proposed charging management strategy is demonstrated for the case of virtually electrified delivery vehicle fleet of a local retail company and virtual electricity production from renewable energy sources, and it is verified against the offline globally optimal mixed integer linear programming benchmark and a baseline dumb-charging scheme in terms of charging cost, renewable energy utilization, and related optimization time execution.