Learning to schedule electric vehicle charging given individual customer preferences

Electric Vehicles (EVs) and their integration in the smart grid are challenges that sustainable societies have to tackle. Large scale uncoordinated EV charging increases peak demand and creates the need for extra grid infrastructure to be covered effectively, which is a costly solution. We propose a decentralized charging strategy for EV customers that offers savings for the individual adopters on their electricity bill and at the same time peak demand reduction, alleviating smart grid from critical strains. We implement our charging strategy through learning agents that act on behalf of EV owners and examine the effect of our strategy under the prism of various market conditions.