Application And Research On The Urban Intelligent Traffic Dynastic Forecasting Model

The important strategy for many big cities to work out solutions to traffic congestion isto develop the intelligent traffic system. At present, the intelligent traffic system has beendeveloped into its intermediate or even advanced stage. The abundant traffic data detectionmeans, already a hot topic in the research and engineering application, provide a fine database for traffic data analysis, control and decision-making. Traditional traffic controlsystem and traffic guidance system are based on real-time traffic flow data detection, oncethe traffic flow status is detected, then the established control and guidance algorithmcalculation is actuated and control implemented, nevertheless the following problems exist:(1) Traffic control and guidance are short of predictability due to insufficient historicaltraffic data analysis and traffic forecasting, and traffic control can only be conducted bychoosing the control plans with the latest short-term traffic parameters.(2) No effective integration of data between the traffic control system and the trafficguidance system causes failure to establish the synergy model for the traffic control andguidance.(3)The existing traffic control system does not take into account the influence of theburst traffic incidents on the model, thus unable to grasp the characteristics of suddenchange of the traffic flow, and traffic control and guidance is obviously lagging behind.In view of the above problems, for improving the predictability of traffic control, itsrapid response capability to the burst traffic incidents, and for the establishment of synergybetween traffic control and guidance, innovative research is carried out in this paper interms of dynamic traffic forecast, traffic incident detection and traffic control and guidancesynergy model. The main research content and its innovation points include the followingfour aspects:(1) Study the traffic flow data preprocessing methods, including the error datadiscrimination and correction method, lost data filling method and redundant data reductionmethod, and apply the methods into Guangzhou Traffic Flow Detection System. Gooddata quality is the basis of the traffic flow forecasting. In practical engineering, traffic flowdata is filled with noise and can’t be used directly. Therefore, the original data must first bepreprocessed, the process is also called data cleaning. The data preprocessing methods summarized and proposed in this article are engineering application oriented, and are ofsignificance in practice guidance.(2) In-depth study is made on the traffic incident detection approaches based on videoimage and these approaches are applied in Guangzhou Traffic Incident Detection System.In this way, the traffic incident detection approaches can overcome dependence on trafficflow detector data analysis, and video detection approaches can make full use of the richcity traffic video resources, which will greatly reduce the project investment and will be ofgreat significance.(3) Put forward a kind of short-time traffic forecasting model adaptable to urbanintelligent traffic control and guidance, including the traffic forecasting model based onSupport Vector Machine (SVM), focus research on the establishment of kernel function andchoice parameters, and verify the model in engineering practice. For improving the trafficcontrol system’s automated adaptive capacity to changes in traffic flow, this paper sets upthe short-time traffic flow forecast mode based on Support Vector Machine (SVM)regression, summarizes the modeling process in engineering application, and uses the datafrom Guangzhou traffic flow detection system to make experiments and quantitativeanalysis against the model for the purpose of verifying the feasibility and validity of themodel and setting the foundation for follow-up engineering application. Finally research ismade on how to optimize choice of parameters with the Particle Swarm Optimization (PSO)algorithm.Ⅳ. Put forward the traffic control and guidance synergy model based on the integratedtraffic dynamics, establish the traffic control and guidance synergy platform based ontraffic forecasting, and finally adopt the simulation methods to verify the feasibility andvalidity of the said model and platform.Presently traffic flow data detection, analysis and control of the traffic control systemand the traffic guidance system are relatively independent of each other. Facing thissituation, this paper first studies the dynamic traffic–oriented information integrationtechnology, and effective information fusion of the short-time traffic forecast inclusive ofhistorical data, traffic incident detection results and real-time traffic flow data is made.Meanwhile research is conducted on the co-ordinated optimization of traffic control andguidance in combination with neural network algorithm. And finally the intelligent controland guidance synergy model with a fusion of dynamic traffic is formed. In the aspect ofsynergy, the model chooses a mode of fusion at decision-making level, proposes a traffic control and guidance synergy model with the Center Co-Ordinate System (CCOS) andadopts the neural network expert system to determine the parameters in the model. Toverify its effectiveness, this paper selected a typical network for simulation, at the sametime used the real data for training in the neural network with expert supervision, and therelevant parameters in the algorithm were derived to implement accordingly traffic controland guidance based on traffic forecasting and real-time traffic status. Lastly the realexample analysis and contrast against the actual traffic flow data before and after theguided control to prove the feasibility and effectiveness of the model.In summary, this paper studies the application of traffic flow forecasting methods, makesin-depth discussions of the traffic flow forecasting model based on support vector machineregression, applies this model, meanwhile taking into account the effect of the burst trafficincidents on the model, to improve the traffic control and guidance synergy model. Theseresearch and practice play an important exemplary role for solving the current hot issue ofthe traffic control and traffic guidance in ITS and can be used as the important andbeneficial references for the follow-up study. The above-said research and practice willprovide important reference value for theoretical research and engineering research.