Context-Aware Optimal Charging Distribution using Deep Reinforcement Learning

The expansion of charging infrastructure and the optimal utilization of existing infrastructure are key influencing factors for the future growth of electric mobility. The main objective of this paper is to present a novel methodology which identifies the necessary stakeholders, processes their contextual information and meets their optimality criteria using a constraint satisfaction strategy. A deep reinforcement learning algorithm is used for optimally distributing the electric vehicle charging resources in a smart-mobility ecosystem. The algorithm performs context-aware, constrained-optimization such that the on-demand requests of each stakeholder, e.g., vehicle owner as end-user, grid-operator, fleet-operator, charging-station service operator, is fulfilled. In the proposed methodology, the system learns from the surrounding environment until the optimal charging resource allocation strategy within the limitations of the system constraints is reached. We look at the concept of optimality from the perspective of multiple stakeholders who participate in the smart-mobility ecosystem. A simple use case is presented in detail. Finally, we discuss the potential to develop this concept further to enable more complex digital interactions between the actors of a smart-mobility eco-system.