Pure Electric Vehicle Battery SOC Estimation And Driving Range Prediction

With the gradual depletion of petroleum resources and the deteriorating natural environment,the battery electric vehicles have become recognized as the leading direction for the future development of the automotive industry due to their environmental protection and low energy consumption.Meanwhile,the research and innovation of its power battery system management technology is a hot topic at present.Battery state of charge(SOC)and vehicle driving range are two important parameters in the battery management system,which are of great significance in guiding the driving of electric vehicles.However,due to the complex and changing driving conditions,the electric vehicle SOC data and energy consumption per mileage are also very different.This has severely affected the accuracy of SOC estimation and range prediction of driving range.In order to improve the accuracy of SOC estimation and driving range prediction,this thesis has conducted some researches,the main contents of which are as follows:Firstly,this thesis briefly introduces the research status of SOC and vehicle driving range,and conducts a simple analysis of the existing methods.Then,in order to obtain relevant data,the model parameters in the automobile simulation software ADVISOR were set according to real vehicle parameters,and the simulation result data was compared with real vehicle test data to verify the rationality of model parameter setting.Secondly,the limitations of artificial intelligence method in estimating SOC are analyzed.Since SOC related data under different working conditions are quite different,if a single data is used for modeling,the generalization ability of the model cannot be guaranteed.However,using mixed data under various working conditions for modeling increases the learning burden of the learning machine and decreases the model accuracy.In order to overcome this defect,a SOC estimation method combining data feature clustering with ensemble learning—— ME-SVR(Multimodal ensemble support vector regression)is proposed.Experiments show that this method can improve the accuracy of the estimation while guaranteeing the generalization ability of the method.Then,in order to establish the vehicle driving range estimation model,the factors affecting the driving range of electric vehicles are analyzed and introduced,and the driving condition is identified as the main factor affecting the driving range of the vehicle.Based on the segmentation of operating conditions and clustering of operating conditions,the SOC reduction value per mileage under four common operating conditions was analyzed.Using the operating condition category and other relevant parameters as inputs,an estimation model of SOC reduction value per unit mileage was established.Finally,in view of the current situation that unknown driving conditions in the future make it difficult to improve the prediction accuracy of driving distance,a battery electric vehicle residual driving distance prediction method based on map information and iterative SVR model is proposed.This method predicts the future driving conditions according to the map information,and inputs the corresponding future driving conditions into the SOC decline value estimation model of unit mileage,calculates the SOC change per kilometer in the future,so as to realize the prediction of the remaining driving range.Based on the actual driving data,the simulation experiment was carried out in ADVISOR,and the experimental results show that the method has a high prediction accuracy of driving range.