Research On Prediction Of Remaining Driving Range Of Pure Electric Vehicles Based On Driving Behavior Analysi

The emergence of new energy vehicles has alleviated the pressure of the energy crisis and can effectively achieve energy conservation and emission reduction in transportation.Under the dual wheel drive of national policies and the market,electric vehicles are increasingly favored by people.However,"mileage anxiety" is still one of the main issues hindering the promotion of electric vehicles.The main reason is that the use of electric vehicles is affected by various factors such as traffic conditions,environmental temperature,driving behavior,etc.,making it difficult to accurately predict the remaining driving range,making it difficult for drivers to make reasonable path planning,reducing consumer confidence in using them.Therefore,improving the accuracy of predicting the remaining range of electric vehicles,alleviating drivers’ "mileage anxiety",and boosting consumers’ confidence in using new energy vehicles have important practical significance for the promotion and popularization of electric vehicles.Firstly,based on the electric vehicle Internet of Vehicles platform,extract the driving message data of electric vehicles,analyze the original message data,filter the analyzed detailed data,and then use the segmented linear interpolation method to process the abnormal data in the table,and expand the data dimension on the basis of the original data dimension to calculate data such as speed,pedal change rate,and energy consumption.In order to facilitate the study of driver behavior habits,the vehicle driving data are divided into driving segments according to three behavioral actions: driving behavior,braking behavior,and sliding behavior.For each segment,parameters related to driving behavior such as vehicle speed,pedal stroke value,acceleration,and energy consumption are calculated.Then,based on the segmentation data of driving segments,the 100 km energy consumption of electric vehicles is taken as the dependent variable,and the characteristic parameters related to driving behavior such as the average value of accelerator pedal travel,the standard deviation of accelerator pedal travel,are taken as the independent variables.The correlation analysis between the independent variables and the dependent variables is performed to extract the driving behavior characterization parameters.Cluster analysis of driving behavior based on driving behavior characterization parameters provides input for predicting the remaining driving range of electric vehicles,and improves the accuracy of predicting the remaining driving range of electric vehicles.In order to guide drivers in energysaving driving,the relationship between the vehicle’s energy consumption per 100 kilometers and its main influencing factors in starting and stopping,different speed intervals,and different behavioral action driving segments was analyzed,and suggestions were made for energy-saving driving to improve the vehicle’s range.Finally,based on driving behavior analysis,driving behavior style recognition is performed on the predicted electric vehicle data.BP neural network model and LSTM neural network model are used respectively.When the number of characteristic variables representing driving behavior is 3 and 4,the remaining driving range prediction model of electric vehicles is trained respectively.The trained model is used to predict the remaining driving range of four electric vehicles used for research,And evaluate the accuracy of the prediction results.The results show that based on driving behavior analysis,using the LSTM neural network model to predict the remaining driving range of electric vehicles,compared to the original vehicle prediction,the average value of the accuracy determination coefficient for predicting the remaining driving range of electric vehicles increased from 0.8539 to 0.9744,with a significant improvement in the accuracy determination coefficient,which is closer to 1.At the same time,research has shown that when the number of driving behavior characterization parameters is 4 or 3,the prediction accuracy of the remaining driving range of electric vehicles is compared.The results show that when the number of driving behavior characterization parameters is 4,the prediction of the remaining driving range of electric vehicles is more accurate,and the four characterization parameters can more fully depict the driver’s driving behavior.