GRU based EV Charging Algorithm for Vehicle-to-Home Applications

Electric Vehicles (EVs) are becoming popular as a sustainable transport solution and the integration of EVs with power systems has emerged as an assuring solution to address the challenges of renewable energy (RE) integration. Vehicle-to-home (V2H) technology has been developed to use EVs as mobile distributed energy storage in residential homes. This article presents a machine learning based methodology to estimate EV charging/discharging power in V2H systems using a gated recurrent unit (GRU) model. The optimal EV charging profiles are generated for the V2H system using optimization and the model is trained on these optimal EV charging profiles. The proposed model is trained and tested using real-life data of time-of-day electricity pricing, load curve of the house and generation from solar PV. The efficacy of the proposed algorithm is evaluated by carrying out a case study to estimate EV charging power in a V2H system using real-life data. The proposed algorithm achieves near-optimal results and no violations in the physical constraints of EV was observed.