Method Of City Traffic Flow Prediction Based On Grey Neural Network

With the rapid development of accelerating of our urbanization process and thenumber of motor vehicles grow rapidly, the traffic problem has become increasinglyprominent. The intelligent transportation system has become the effective way to solvecity traffic problems. Traffic flow forecasting is one of the key techniques in intelligenttransportation system. And it is mainly used for prediction and analysis of urbannetwork traffic flow state of each node and route. The accuracy of traffic flowprediction is a prerequisite for the realization of city traffic flow control.The traffic flow is a nonlinear system with a multivariable, time-varying andcomplex structure. The traditional single prediction model can only summarize parts ofgeneral characteristics of the system, and the prediction precision is limited. For thatreason, this paper presents a city traffic flow combination prediction method based onthe analysis of related intersections and grey neural network. This paper has a detailedstudy on the traffic flow prediction model, the method and its implementation. Themain work includes the following aspects:(1)This paper puts forward the systematic method of traffic data preprocessingand the judgment method and process of data errors, missing and redundant is putforward. It is effective to remove the interference of noise data and reduce the dataredundancy. It also improves the efficiency and accuracy of the follow-up traffic flowprediction.(2)Combining the respective advantages of grey system and neural networktheory, this paper sets up the grey neural network model. To predict the city roadtraffic flow by using this model, the simulation results show that it can effectively improve accuracy and real-time of the prediction.(3)This paper puts forward one method of road intersection traffic correlationanalysis. And use the historical data to conduct the flow correlation analysis for roadnetwork target intersection through the principal component analysis method. And itsets up a combination forecasting model to predict the missing intersection trafficflow data by using the network related intersection data.This paper takes the urban road intersection as a unit, and according to the actualconditions, takes the flow and interval length between intersections into account,establishes the local area network for part of the administrative region intersections ofShenyang city. The article makes a detailed study on the prediction model, method andits implementation, simulates and validates the real-time and accuracy by using thehistorical data of relevant junctions and target intersections.