Study On The Urban Traffic Flow Dynamic Route Guidance Base On The RFID Data

The rapid development of urbanization brings the growing demand and the pressure to the existing urban traffic system,traffic jams have been the persistent ailment of the city’s economic development.There are serious problems such as environmental pollution,energy waste due to transportation efficiency declines,traffic jams should be urgently solved.And the urban traffic guidance system based on the intelligent data collectors plays an important role while relieving the traffic jams.In this paper,traffic data characteristics collected from the radio frequency identification devices(RFID)environment were studied at first,and methods to extract traffic characteristics were built,and the traffic simulation framework was built based on RFID data,and then the traffic flow short-term forecasting method was built based on RFID data,the traffic state fuzzy classification research was carried out based on RFID data and the dynamic traffic guidance path research was completed based on the traffic state classification.In this paper,methods of collecting vehicles data with RFID were firstly introduced,considering that the most RFID collectors are set in the lane section of the intersections downstream,and the single section records are statically.In order to match the vehicles complete route with the RFID data and dynamic demand,RFID collector spatial locations should be matched with the vehicles records while considering the intersection time schedule.Based on the data quality analy’sis,methods of extracting traffic characteristics based on RFID data were built in this paper,concepts of traffic engineering were used as the key factors,and the plate number column and passing time column were applied to extract the parameters of single section and adjacent sections,traffic volume,speed,travel time,time headway,traffic flow directions were respectively extracted.Statistic approaches were used to analyzing the parameters characteristics,which would be used as the input factors for building the vehicles simulation frameworks.Vehicle lane-traveling characteristic were firstly analyzed to build vehicles micro simulation frameworks.Vehicles travel in the sections from the upstream detectors to the downstream detectors,and the characteristic analysis results were combined with the linear car-following models and safety lane-change models to build lane-traveling rules.Meanwhile,the time of arrival at the stop line and the signal running schedule were used to decide whether vehicle stop or not,the results were used to accomplish the simulation framework with the lane-traveling rules.When analyzing the examples,random traffic parameters were produced by the traffic parameters statistical models in MATLAB,and the vehicles were simulated in the single intersections including RFID detectors.Then the simulation results were discussed,and travel time,queue length can be directly acquired.Simulation environment can be applied in extracting the dynamic traffic parameters from the static records and selecting the parameters used for traffic state identification.To further get the traffic dynamic characteristic from the static records,traffic short term forecasting models were developed based on the RFID data.Considering that the random signal characteristic can be found in the traffic flow time series data,the wavelet methods were used to obtain the approximate and details variables.Then the methods of support vector machine(SVM)were applied to build the forecasting model,with the GA,PSO optimization methods.In the respects of values precision and computational efficiency,numerical experiments were used to decide the numbers of the input factors.In the speed forecasting examples,results of 2min interval and 5min interval were respectively implemented,and the precision match the demand.The forecasting methods were also used to the threshold de-noising data,and results indicated that the methods would be used to dynamically forecast the RFID static records.Forecast methods could be used to analyze the static traffic flow parameters extracted from the RFID records.However,dynamic traffic state characteristic should be analyzed by the different traffic parameters.Based on the vehicles micro simulation framework,a method of traffic state classification by the mean travel distance delay,mean travel speed and queue length was developed.The methods were combined with the fuzzy synthetic evaluation method,and the traffic parameters extracted from the simulation were treated as the input factors.Trapezoidal membership functions were built and the traffic state of contiguous detector sections were classified including left,straight,and right directions of the intersections,and the relationship between weighting matrix and results value were analyzed.Traffic dynamical characteristic were further discussed by the developed methods,and the results could be used to search the urban traffic guidance route.Route guidance is an important function of the traffic flow guidance system.In this paper,a dynamic route searching algorithm based on the urban traffic state were developed.Local network was selected as a study object,data from the vehicles micro simulation were used to extract the traffic parameters,and the mean travel distance delay,travel speed,and the queue length were used to classify the traffic state..Then the traffic state evaluation results were treated as the directed arc dynamic weight values,and the Manhattan distance were used as the heuristic function and the heuristic search algorithm was chosen as the shortest path search algorithm.Shortest routes were found for traffic guidance for the local traffic network,and the results could be used to direct the guidance route for the traffic management system based on the RFID data.