Adaptive Multi-stage Optimisation for EV Charging Integration into Smart Grid Control

The increasing amount of both, renewable energy production and electric vehicle usage, puts considerable stress on smart grids, making it necessary to synchronize vehicle charging with energy production, but also allowing to use those vehicles as additional energy storages. In this paper, we combine machine learning and evolutionary algorithms to create near-optimal vehicle charging schedules from incomplete information. Using multi-agent systems and process modelling techniques, the different stages can easily be combined and distributed. The result is a reusable and extensible solution that is used for optimizing charging schedules in many different project settings.