The paper deals with a novel methodology for calculating the optimal use of existing charging infrastructure in the electric mobility sector. Volatile renewable energies should thus be able to be better used by an intelligent IT infrastructure. By means of reinforcement learning, stakeholders are identified, their contextual information such as time and place is processed and then optimally guided. In this way, the needs of each stakeholder can be met, which can be vehicle owners, operators of charging stations, network or fleet operators, for example.
Reinforcement Learning stands for a whole range of methods of machine learning. Here, a so-called software agent independently learns a strategy to maximise the rewards he receives. Rewards are positive or negative feedback on the actions of the software agent. The models of reinforcement learning try to imitate the learning behaviour in nature.
In the methodology described, the system learns from the environment until the optimal allocation strategy for loading resources within the system boundaries is achieved. The concept of optimality is considered from the perspective of several actors involved in the intelligent mobility ecosystem and will be further developed to enable more complex digital interactions between the actors of an intelligent mobility ecosystem.