| The rapid development of human society,science,technology and economy,cannot be separated from the contribution of fossil energy such as petroleum,but as it continues to be consumed,it is non-renewable.In addition,the environmental pollution caused by the extensive use of fossil energy has also attracted the attention of governments all over the world.Therefore,in order to solve the problems of environmental pollution and insufficient resources,people are vigorously developing the electric vehicle industry,and the current development of electric vehicles can be described as changing with each passing day.As large-scale electric vehicles enter our lives,new problems will inevitably appear.As a new type of load connected to the grid,it increases the burden on the grid and also brings many uncertainties to the security of the grid.Therefore,it is of great practical significance to conduct research on the prediction of the charging load of electric vehicles and the orderly control and guidance of charging and discharging.First,analyze the factors that affect the charging of electric vehicles,including electric vehicle types,user habits,vehicle charging modes,and battery types.Combine the actual load data of a charging station in Hangzhou for preprocessing and standardization,for the establishment of load forecasting models Foundation;Secondly,the random forest is combined with the convolutional neural network,and the RF-CNN prediction model is used to predict the charging load of the electric vehicle charging station.In order to study the accuracy of the prediction model.compared with the three models of BP,SVR and LSTM,the RF-CNN model has the smallest loss value during the training process of the data,keep below 3%and the loss fluctuation is the most gentle.At the same time,the prediction results of each model are compared with the actual load curve.The load curve predicted by RF-CNN is the closest to the actual situation,and the prediction accuracy is the highest,and the error level is maintained at about 5%,which is much lower than the error values of the other three models;Finally,in order to further study the orderly charging strategy of electric vehicles,an orderly charging control model is constructed using a two-layer optimization control method.It is expected to guide and control the charging period of electric vehicles,and the particle swarm algorithm is used to solve the model.In the case of orderly charging.after the orderly guidance of the vehicle,the peak-to-valley difference of the charging load has been reduced by 26.7%.Under the current local peak-to-valley price policy,the user’s peak-hour charging fee has dropped by 39.35%.and the average time has dropped 6.05%,the overall charging cost has been reduced by 14.07%.The experimental results show that the RF-CNN model can complete the charging load prediction task more accurately under the condition of lower training error;for the orderly charging guidance control of electric vehicles,the goal of peak load transfer is achieved and the load is reduced.The peak-to-valley difference avoids the situation of peak summation with the basic load of the local power grid,ensures the safe operation of the power grid,reduces charging costs,and brings economic benefits to users. |
