Research On Power Battery SOC And Driving Range Estimation Of Battery Electric Vehicle

Energy sustainable development and application had become a major issue. Electricvehicle technology is important strategy of sustainable development energy，now allcountries in the world have been put more effort in develop electric vehicle technology.Battery technology is a core technology in electric vehicle and state of charge (SOC)estimation is an important content. Driving range calculation was base on accuracy SOC;both have important significance to electric vehicle driving.This paper did research on the power battery SOC estimation and driving rangecalculates. Firstly, LiFePO4Li-ion battery characteristics were analysis by experimentand an equivalent circuit model of the battery was established. The SOC of LiFePO4Li-ion battery was estimated by ANFIS. Two inputs ANFIS prediction model and threeinputs ANFIS prediction model were established respectively and a hybrid algorithmcombined with experimental data to study the model. The result of instance shows thatthe ANFIS can accurately predict the LiFePO4Li-ion battery SOC values, and the resultby three inputs ANFIS model was better.Then the ANFIS model estimation SOC was used for driving range calculation.Taking a battery electric vehicle as an example, the established vehicle simulationmodel based on Matlab/Simulink was used to simulate its remaining driving range, andthe calculated values are compared with the actual values, error below5%.Finally, a driving cycle identification method, which combined principal componentanalysis with fuzzy clustering analysis, is proposed to estimate the remaining driverange of pure electric vehicle. Twenty representative driving cycles were selected anddivide these cycle data into215cycle segments. Choose12characteristic parameters toanalyze these segments by principal component, fuzzy C-means clustering and drivingcycle identification. Using vehicle model to solve driving cycle identification, energyconsumption calculation and remaining drive range estimation. A real battery electricvehicle was tested on drum test bench under ECE15cycle. Compared remaining driverange estimated value with test value, the result shows that mean absolute error is0.742km, the average percentage error less than3%.