Description
Reinforcement learning, a machine learning technique which is powerful for solving optimization problem, will be adopted for developing EV fleet smart charging controls. The degradation of EV batteries will be investigated under different smart charging strategies. A C2V2C business model will be designed for optimizing the PV power trading between the Swedish building communities and EVs. An optimization based C2V2C power trading tool will be developed in Matlab environment. Considering the potential impacts of future climate change on the PV power production, building electricity demand and EV battery degradation, this project will also assess the impacts of the future climate change on techno-economic performance of the C2V2C model in the Swedish context. Three specific results are expected: (i) a C2V2C control optimization method, (ii) a C2V2C power trading tool for the building communities and EVs, and (iii) resilience of the C2V2C model under the future Swedish climate scenario.
This project can provide knowledge for the stakeholders to understand the potentials of EVs for power regulation in building communities and design efficient, economical and resilient C2V2C framework. Such C2V2C framework can bring significant benefits to various stakeholder: (i) The building communities can become more self-sufficient and thus less dependent on the public power grid due to the enhanced local power demand-supply balance; (ii) The EV owners can get economic incentives from participation in the power regulation activities; (iii) Due to the reduced large peak power demand and PV power exports, the power grid can increase the hosting capacity of PV and EVs without expensive upgrading of the existing electricity distribution systems.