Probabilistic Forecasting Model for Non-normally Distributed EV Charging Demand

A method for probabilistic electric vehicle (EV) demand forecasting is proposed in this paper. The EV demand in a certain area is forecasted by an ensemble forecasting model. The forecast result includes a deterministic forecasting and prediction interval that indicates the probability of deviation from deterministic forecasting. In the case study, an actual observed dataset from the UK is used to verify the proposed algorithm. The results show that the target EV charging demand to be forecasted shows 80% prediction interval coverage for 27 days out of 30 simulated days.