Forecasting The Driving Range Of Pure Electric Vehicles Based On RBF Neural Network

With the worsening of the global climate. In recent years, motor vehicle’s energy conservation and emission reduction has been widely discussed. Among the discussions, promotion of new energy vehicles is the most effective one. Therefore, a wave which focuses on researching and creating new energy vehicles swept the world.Among all the new energy vehicles, pure electric vehicle is the one that produces minimum pollution, and almost make no destruction on the environment. However, this kind of vehicle costs too much money. In addition to the cost of the locomotive itself, the battery for the vehicle spends too much. Moreover, the battery’s service capability and service life are easily influenced by working conditions. Therefore, during the running period of pure electric vehicle, it’s necessary to monitor battery’s condition as well as the vehicle’s operation and forecast pure electric vehicle’s driving range.In order to achieve the requirements above, this thesis designs the operation and monitoring system of pure electric vehicles, and describes the system design ideas and the data flow. Then introduces the key technology used in the database and the communication protocol, which supports the electric vehicle data management system and the GIS monitoring system.The programming soft ware is visual studio2008, the development language is C#, the database is SQL SEVER2008.This thesis takes the operation and monitoring system of pure electric vehicles in Guangzhou Asian Games for example. Details the deployment of the battery set on the bus. According to the battery work factors:1) designes the arithmetic for data processing and makes it real.2) Analyzes the processed data, and determines the main factors which affect the mileage.3) Trains the the Radial Basis Function (RBF) Neural Network designed in this thesis, and obtains the RBF Neural Network Model which can forecast the pure electric vehicle’s driving range.4) Through revising and analyzing the model, the practicability of the model is proved.