Optimal charging/discharging of grid-enabled electric vehicles for predictability enhancement of PV generation

This paper proposes a collaborative strategy between the photovoltaic (PV) participants and electric vehicle (EV) owners to reduce the forecast uncertainties and improve the predictability of PV power. The PV generation is predicted using an auto regressive moving average (ARMA) time series model. Fuzzy C-means (FCM) clustering is used to group the EVs into fleets with similar daily driving patterns. Uncertainties of the PV power and stochastic nature of driving patterns are characterized by a Monte Carlo simulation (MCS) technique. A particle swarm optimization (PSO) algorithm is developed to optimally use the vehicle-to-grid (V2G) capacities of EVs and minimize the penalty cost for PV power imbalances between the predicted power and actual output. The proposed method provides a coordinated charging/discharging scheme to realize the full potential of V2G services and increase the revenues and incentives for both PV producers and EV drivers. An economic model is developed to include the V2G expenses and revenues to provide a complete picture of the cost-benefit analysis. The proposed model is used to evaluate the economic feasibility of V2G services for PV power integration.