Short-term Prediction For Congested Traffic Flow Based On Chaos Theory And Wavelet Neural Network

In recent years,traffic congestion problems have received considerable attention of relevant traffic management departments.Some advanced control concepts have provided novel and effective solutions to these problems.However the currents studies on the short-term prediction of traffic flow mainly focus on the short-term prediction in the free traffic flow,and seldom consider the traffic flow operation characteristics under congested condition,on the other hand,most of the current studies on prediction of traffic flow are based on the simulation verification of measured data and MATLAB,and there are few short-term prediction studies combining professional traffic simulation software to realize traffic flow.Therefore,the short-term prediction of congested traffic flow based on chaos theory and wavelet neural network in this study is of certain theoretical and practical significance to traffic management and control.This thesis studies the traffic flow operation characteristics under congested condition based on the various states of urban expressway traffic flow and uses the wavelet neural network for the short-term prediction of congested traffic flow,in addition,realizes the simulation application of the short-term prediction of traffic flow by using the professional traffic simulation software TransModeler.The specific contents include: firstly,based on the operation characteristics of urban expressway,this study analyzes and divides the operating state of the traffic flow of the urban expressway,and obtains the measured data of the congested traffic flow;secondly,the thesis analyze the chaos characteristics of parameter time series and conducts phase space reconstruction using the measured data of congested traffic flow;then,achieving the short-term prediction of congested traffic flow based on wavelet neural network;finally,builds a typical urban road network through the software TransModeler,and achieves the simulation application of the short-term prediction of traffic flow based on TransModeler secondary development simulation display platform of the short-term prediction of traffic flow.The main work of this thesis is as follows:1.Studies on operating characteristics of congested traffic flow based on threephase traffic flow theoryConsidering the difference between the operating characteristics of the traffic flow of urban expressway and the highway,this thesis analyzes the operating characteristics of the congested traffic flow of the urban expressway in various states based on threephase traffic flow theory,and verifies the traffic flow by the measured data of traffic flow,then selects congested traffic flow time series to preprocess,finally,obtains the preprocessed congested traffic flow time series.2.Analysis of congested traffic flow time series based on chaos theory and phase space reconstruction.Considering the nonlinearity of the congested traffic flow,this thesis uses the maximum Lyapunov exponent to determine the chaotic characteristics of the time series,and conducts numerical simulation verification based on the measured time series of the congested traffic flow;then,analyzes the intrinsic characteristics of the traffic flow time series by using the phase space reconstruction theory;finally,the phase space reconstruction of the measured congested traffic flow time series is performed.3.Short-term prediction of congested traffic flow based on wavelet neural networkConsidering the particularity of the congested traffic flow,this thesis builds the short-term prediction model of congested traffic flow based on the wavelet neural network,and analyzes of prediction performance using numerical simulation.Firstly,with the theoretical knowledge of wavelet neural and the study on the adjustment methods of parameters,this thesis selects the average speed time series of traffic flow and uses the short-term prediction model to obtain the prediction performance;secondly,proposes an improved short-term traffic flow prediction model based on genetic algorithms aiming at the shortcoming of the original prediction model;finally,analyzes the short-term prediction of average speed time series based on the improved wavelet neural prediction model,and evaluates the performance of the improved model.4.The simulation verification for the short-term prediction of traffic flow based on TransModeler.Considering the practicality of short-term prediction of traffic flow,this thesis builds the typical urban expressway network through the professional traffic simulation software TransModeler and realizes the short-term prediction simulation of the traffic flow of the urban expressway combing with the algorithm of short-term prediction of traffic flow.First,constructs the typical urban expressway network and adjusts the parameters of the simulation to make the built expressway network show the operating characteristics of traffic flow of the urban expressway;secondly,embeds the algorithm of the short-term prediction traffic flow to the TransModeler platform based on the secondary development;finally,realizes the simulation of short-term prediction of traffic flow based on TransModeler and analyzes the short-term prediction of traffic flow based on TransModeler and analyzes the short-term traffic flow using the TransModeler simulation data of urban expressway network.