Research On Algorithms Of Traffic Prediction And Route Guidance Based On Floating Car GPS Data

Dynamic traffic prediction and guidance model are important contents of Intelligence Transportation System.At present datum of dynamic traffic prediction and guidance model are often obtained by fixed detectors(such as line detectors,super voice detectors,video frequency detector...etc.).These equipments have some shortcomings,such as installed and supported highly,overlay scope sally,only fix position data.So these equipments can't satisfy the need of ITS development.Because GPS,GIS and wireless communication technique are extensively applied,the float car installed GPS collect dynamic traffic data and become the important development direction of collecting data.The float car has some advantages,such as constructible period shortly,time strongly,overlay scope largely,data accuracy highly and so on.So it is important that dynamic traffic prediction and guidance model of the FCD are researched in theory and technique.In paper,matching GPS and GIS datum,predicting the travel time and traffic guidance model are researched.The corresponding theory method and mathematics model,including the angle-distance matching model and the city dynamic traffic prediction model by long interval,are creatively improved.The kalman smooth deal model and relevant adaptive filtering model are first introduced into the dynamic traffic prediction. The GPS datum of FCD who have long interval is the first to be applied into dynamic traffic of city.The some problems that datum compromise of GIS and GPS,large error of dynamic travel prediction and realizing guidance model,are solved that has existed in dynamic traffic prediction by the FCD in paper.Finally the next research content is put forward.The situation about dynamic traffic predication and induced model is introduced by the FCD in home and aboard.The theories,including kalman,relevant adaptive filtering,dynamic OD,optimal path,are researched.Through the problem is analyzed in current research achievement,such as highly mistaken matching rate of GIS and GPS,low accuracy of travel dynamic prediction,difficult express of city traffic information in network.Then the content of the paper is determined.For the problem that the GPS point of floating car cannot fall on the road network,data filtering,coordinate converting and fusing GPS and GIS are researched,the angle-distance matching model is the first to put forward.When the sampling period of the floating car is 1 second,the model can avoid beating in parallel and the wrong matching at the crossroads.At the same time the model has the speed of the point-to-line matching and the accuracy of the line -to- line matching.For avoiding the problems that the speed of the the whole road by the instantaneous speed of the float car and predicting change information with time over,the kalman theory is researched.The method of building kalman filter equations is the first to be improved basing on the long interval GPS datum in city traffic,and the parameter regression smooth deal is used on this basis of the results of Kalman filter.For the affect of travel time from system noise and observation noise,the theoy of relevant adaptive filtering is first introduced into the dynamic traffic prediction.Therefore the method of building the equation is put forward by the float GPS data.The error is cooperated by relative error,relative error and mean square root of the square,the equalization coefficient etc.The result shows that kalman's accuracy is higher than that of relevant adaptive filtering.For expressing traffic information,the theory of building network is researched in GIS, the improving method is put forward by fixing code rules.The one-way traffic,turning restrictions,traffic control are implemented in geometry network by strict code rules.Then the improving network is suitable for the demand of the dynamic traffic-induced.For dynamic optimal path problem,the dynamic OD theory is researched.The improving Dijkstra method is put forward.The Dijkstra algorithm is improved for meeting the need of the optimal path in the traffic induced by the theory of Dijkstra shortest path and the road network model builded of the paper.The theory and methods of dynamic traffic prediction are systematically researched based on the FCD in the paper.Especially the new research result is obtained in the GPS matching, the travel time prediction based on the Kalman filter,building road network model,the search method of optimal path,and provides theoretical and technical support for the practical application of the FCD.