Energy Consumption Prediction Of Electric Vehicles Based On Combination Model

Nowadays,the promotion of electric vehicles in China faces two major problems: first,the driving range is lower than that of traditional fuel vehicles at the same price;second,the state subsidies for new energy vehicles show a downward trend after 2019.Accurate prediction of electric vehicle energy consumption,on the one hand,can accurately estimate the driving range,reduce the driver’s mileage anxiety,on the other hand,can correctly evaluate the use cost of electric vehicles,and help consumers make the right choice.Therefore,it is of great practical significance to predict the energy consumption of electric vehicles.First of all,in order to obtain sufficient real vehicle data,this paper builds a electric vehicle data acquisition system composed of vehicle information acquisition terminal and epigynous machine.Vehicle information acquisition termina collects,processes and sends vehicle operation data.The epigynous machine is responsible for the collection,analysis and storage of vehicle data.The electric data acquisition system monitors 192 vehicles from four cities in real time for two consecutive years,and obtains sufficient real vehicle data,which provides data basis for accurate prediction of energy consumption.Then,this paper systematically analyzes the energy consumption direction of electric vehicles,mainly including driving resistance energy consumption,braking recovery energy and accessory energy consumption.The fixed step division method is used to divide the kinematic segments,and the actual energy consumption in the segments is calculated by collecting the voltage and current data of the battery,which is used as the comparison data of the energy consumption prediction model.Finally,this paper constructs a single regression prediction model and a combination prediction model of regression analysis and radial basis function(RBF)neural network.Based on the energy flow model,considering the influence of temperature on the vehicle energy consumption,a multiple linear regression analysis model is established,and the parameters of the regression model are identified by using the data collected from real vehicles.The nonlinear prediction error of the regression model is obtained by comparing the predicted energy consumption with the actual energy consumption.The travel temperature is taken as the input node of RBF neural network model,and the nonlinear error of multiple linear regression model is taken as the output node.The RBF neural network model is trained by real vehicle data to compensate the nonlinear error and improve the prediction accuracy.The final combined energy consumption prediction model is the combination of linear prediction part of regression model and RBF neural network compensation for nonlinear error compensation.Compared with the prediction results,the prediction accuracy of the combined model of multiple linear regression and RBF neural network constructed in this paper is higher than that of the single multiple linear regression model,which can improve the problem of insufficient prediction accuracy of the single prediction model,and has engineering application significance for estimating the driving range of electric vehicles and evaluating the economy.