Evaluating The Impact of Load Forecasting Error on Scheduling Performance of EV Smart Charging

With the rapid promotion of electric vehicles (EVs), smart charging has drawn increasing attention. By shifting EV charging loads from power peak to valley periods, smart charging is believed to be able to improve the safety and stability of the grid operation. However, the uncertainty of future electricity load on the power grid poses challenges for implementing smart charging effectively. Despite extensive research aimed at improving the accuracy of electricity load forecasting, errors remain inevitable. In this paper, the impact of load forecasting error on the scheduling performance of electric vehicles (EVs) is comprehensively analyzed, considering varied load patterns and EV numbers. Firstly, electricity load data is obtained from a university, and an EV charging behavior dataset is established using statistical data. Subsequently, based on the actual data, electricity load profiles with varied error levels are generated by employing the mean absolute percentage error (MAPE) as evaluation criteria, which is assumed to result from day-ahead load forecasting. Finally, case studies demonstrate that MAPE should be less than 2% to ensure the scheduling performance degradation rate (SPDR) does not exceed 5% under varied load patterns and EV numbers.