Research On Forecasting Of Traffic Condition Based On Chaos Theory

It is now recognized that the intelligent traffic management and control is the bestway to suppress traffic jam and to improve road resources utilization. While, theaccurate prediction of traffic condition is the prerequisite and basis to achieve guidanceand control of traffic flow, and it is also the key to realize the transition of intelligenttransportation system from passive reaction of traffic condition to actively reaction.The traffic system, which is composed of verious micro individuals, is a complexsystem, and there is a complex and nonlinear relationship among these microindividuals. Moreover, it is disturbed by several of uncertainties, such as individualbehavior of driver, traffic incident and surroundings, which make the traffic conditionshow high correlation and volatility. While, the chaos theory characterized by non-linearis known as the20th century’s third natural scientific revolution, and the theory make itpossible to describe the evolutional characteristics of complicated traffic system.Therefore, this paper studied on the forecasting of traffic condition time-series based onchaos theory.A forecasting algorithm for traffic condition is introduced by using the multipleparameters fusion and phase space reconstruction theory in this thesis based on theassociation relationship among the traffic parameters. The modeling and prediction oftraffic condition was emphatically researched on terms of single-parameter andmulti-parameters according to chaos theory. The main contents of the dissertation are asfollows:①The method for traffic condition estimation is improved by introducing newimpact factors, and a method of difference data preprocessing was proposed to ensurethe real-time estimation and prediction of traffic condition.Based on analsis of detection methods of traffic condition and data preprocessing,taking time varying characteristic into account, a differential data preprocessing methodfor time series was proposed in order to enhance the efficiency of data transimission,storage and computing. The traffic parameters time series were processed by using thedata type transforming and different transmission, and the GPS measured data wastaken as example to check the technology of data preprocessing proposed in this paper.On this foundation, by referencing the fuzzy theory, the impact factors (road condition,split green ratio and road grade) were introduced to identify traffic condition, and on the basis of a summary of previous study, an improved discrimination method of multifactor model was put forward based on the road average speeds in this paper.②A traffic condition chaotic forecasting model was proposed based on the roadrelative average speed and road free flow velocity in order to improve the effect ofvarious road conditions on the evaluation result when merely using average speed toevaluate traffic condition and to raise the accuracy of prediction of traffic condition.As one of the basic traffic parameters, the road average speed supplied the moredirect evidence to identify road traffic flow condition. However, when merely usingroad average speed to describe traffic condition, it didn’t be taken into account that therelative road condition information among different road segments, so the accuracy ofprediction of relative traffic condition would not be assured in different road segmentsin road network. On the other hand, the traffic system was chaotic in nature. Therefore,it was necessary to analyze the chaotic characteristic of relative road average speed indifferent road segments to improve the effect of various road conditions on theevaluation result when merely using average speed to evaluate traffic condition and toraise the accuracy of prediction model of traffic condition. Finally, according toanalyzing measured data of average speeds in different roads, it was proved that theprediction model could be improved and the prediction accuracy be enhanced byintroducing the new traffic parameter in traffic condition forecasting.③A chaos prediction model, which introduced the detail fusion algothm ofmulti-parameters and method of prediction, was proposed based on relationship ofdifferent traffic parameters.Generally, the existing forecasting models were supported by single parameter,however, because of the highly complexity of the traffic system, the forecastingprecision was lower due to lack of the complete traffic information. Therefore, thetraffic condition was depicted from vary point of view based on multi-parameters,aiming at enhancing the accuracy of prediction by using the more complete trafficinformation.In the finally part, the summary to the research works of the full text is given, andthe direction of further study was pointed out about traffic condition chaoticcharacteristics analysis and prediction.The study work and the main achievements of this paper have a significant value indata fusion and chaos theory.