Research On Public Intelligent Transportation Information Service System Based On Neural Network

With the development of urbanization,modernization and the process of motorization,pressure of urban traffic is increasing,some regions of big cities has more and more serious traffic congestion,which bought a huge economic waste of resources and serious environmental pollution.so the importance of development of Intelligent Transportation Systems(ITS) is recognized. The public intelligent transportation information service system thoughted as an important research in the field of ITS, is considered to be an effective way to ease the traffic pressure.This thesis complies with the current issue, provides an overview of the situation of development of the public intelligent transportation information service system, A basic framework of public transportation information service system is built up,because the prediction of travel time is the most important research in the public intelligent transportation information service system,the predictive information directly relates to the relevant sub-function of the public intelligent transportation information service system whether can effectively work or not,So this paper focuses on several key technical points of how to predict travel time effectively.In the paper, firstly, the implementation and application of the public intelligent transportation information service system at home and abroad are discussed,the function of subsystems are analyzed, Including traffic flow information collection and processing system, vehicle location system, travel management system and traffic control and safety system.Then,data collection and preprocessing of bus based GPS vehicle navigation system are theoretical analyzed, GPS data only reflects the instantaneous position of vehicles,so, if the raw data is directly used in the prediction of travel time,which will produce large errors.Therefore this paper discuses the method of getting interval average vehicle speed based on distance and time,how to select valid data from the collected results,replace the instantaneous speed by processed average speed in order to calculate the travel time.more effectively.According to the characters of data of bus and traffic flow, models and methods of travel time prediction are discussed,the model of travel time based on prediction BP neural network is built up with discussing determination and selection of model parameters. In the condition of expand the size of input to improve the prediction accuracy, present a model based on the combination of genetic algorithms and artificial neural network, because BP neural network algorithm is improved by adding genetic algorithms to learn the parameter of BP neural network.This paper compares simulation results verify this improvement could increase the accuracy and efficiency of the model.