Modeling And Analyzing Of Driving Mileage Of Electric Vehicle Based On Data Driven

With the growing of urban environmental pollution and the development of new energy vehicle technology, the electric vehicles attract people’s attention with its "zero pollution" property. But due to the short range of electric vehicle (EV), the drivers may more worry about the driving mileage, which severely restricts the popularization of EVs. However in the real road network, because of the real-time changes of traffic environment and traffic condition, there are many differences in the driving mileage of EVs between the real world and the present research which not give application of EVs better guidance. Therefore based on the data of the actual operational EVs, it is very necessary and meaningful for developing EVs to predict the driving mileage.Most previous research use the experimental simulation data, different from that, the lithium battery data of battery electric vehicles (BEVs) which are used in this paper are collected from the logistics EVs produced by BAIC BJEV. As the working environment of batteries is more complex, the real-time collected data are more meaningful and practical than the laboratory simulation data. In order to guarantee integrity and accuracy of data, before proposing the driving mileage model, the paper needs to delete wrong or repeat data from the numerous of battery data. The battery data are processed by some interpolation methods to become continuous completely. Then, the paper deals with averaging process to make battery data more standardized. These processed data are used in the data analysis and the development of model.In order to accurately predict the driving mileage, the conditions of the vehicle and road traffic are considered in this paper. Combined with the real-time operational data, the analysis of the driving mileage is presented based on data drive modeling theory. This paper first deeply analyzes the relationships among the collected data, and then introduces the percentage of electricity consumption per kilometer. Analysis shows there is obvious negative correlation between SOC and the driving mileage. And there is the nonlinear relationship between speed and the percentage of electricity consumption per kilometer. Thus the paper first develops a linear model combining SOC with the driving mileage and model results show that the errors are much large Then the speed factor is brought in to adjust the model, the validation experiments show that model error is associated with the temperature. So, we added an adjusted factor corresponding to the temperature to adjust model again, the validation experiments show that the adjusted model based on data drive modeling theory has a very high performance, which can satisfy the actual needs.The paper applies the driving mileage model to predict the residual mileage of EVs and calculates the economic speed and the most comfortable operational temperature of EVs. Finally the paper summaries the suitable conditions for the driving mileage model, which has a favorable guidance to drivers.