Research On Short-term Load Forecasting For Electric Vehicle Charging Stations

In recent years,due to the emissions from fossil energy use by humans,the global environment is deteriorating,global warming is intensifying,and energy shortages are becoming more and more serious.The promotion of new energy vehicles,especially electric vehicles(EVs),has become an important measure to change this situation.Compared with fuel vehicles,EVs use electricity instead of directly consuming fossil energy,which can be generated by hydropower,nuclear power and renewable energy,thereby significantly reducing environmental pollution and reducing dependence on petroleum resources.However,the largescale EV charging station load will impact the power grid and affect its stability.Therefore,the load forecasting of EV charging stations is an important direction to be explored in the future,which is of great significance for improving the safety and stability of future power grid operations.In this paper,the operation characteristics of EV charging station are studied in depth,and the short-term load forecasting model of charging station is proposed.Based on the analysis of prediction error,the method of eliminating the influence of error by establishing hybrid energy storage system(HESS)is proposed.The specific research content is divided into three parts: The first part is the traffic flow prediction based on the WT-ARMA-CNN combination model.The wavelet decomposition(WT)is used to decompose the traffic flow sequence into time series of different frequency components.Then,the stationarity test is performed on each frequency component.The stationary series is predicted by the Auto-Regressive Moving Average(ARMA)model,and the non-stationary series is predicted by Convolutional Neural Networks(CNN).Finally,the prediction values of ARMA model and CNN are reconstructed to obtain the final traffic flow prediction value.The second part is the charging station probability load modeling based on data mining and user behavior.Using advanced data mining technology to obtain the EV arrival rate of the charging station,and then using the queuing theory model considering user behavior and capacity limitation to realize the conversion of traffic flow to charging load.The third part is the capacity configuration of hybrid energy storage system based on signal processing technology.By analyzing the prediction error,a hybrid energy storage system is established to eliminate the influence of errors.The empirical mode decomposition(EMD)is used to decompose the EV charging load prediction error to obtain different frequency components,and then the capacity of the hybrid energy storage system is configured according to the characteristics of different types of energy storage devices.Simulation results show: The proposed WT-ARMA-CNN traffic flow prediction method has higher prediction accuracy and better performance than single CNN,WT-ARMA-BPNN and WT-ARMA-SAE.It proves the superiority and feasibility of the model.The predicted load of the charging station is not much different from the real load data.It also verifies the effectiveness of the proposed load modeling method and can realize the transformation from forecasting traffic flow to predicting charging load.Finally,the results of the capacity allocation of the hybrid energy storage system show that the capacity scheme can better deal with the impact of load forecasting error,so that the predicted load can be applied to the actual power transmission planning and scheduling.The effectiveness of the capacity allocation method of the hybrid energy storage system is also proved.