The Study On Algorithms For The OD Matrix Extimation Based On The Turning Flow Model

The dynamic OD matrix estimation has been the problem that the scholars hope earnestly to resolve. The OD matrix acquired by the traditional methods of traffic survey doesn't have real-time feature, and it can't satisfy the requirement of real-time traffic management, and besides, because of huge investment, it is hard to promise that GPS and roadside system will cover all of the road and that each car installs a vehicle system in the short period by the methods of installing GPS vehicle system to acquire the dynamic OD matrix. Therefore, the scholars have always been looking for the methods of calculating dynamic OD matrix through observable traffic flow.This paper analyses the domestic and abroad methods of calculating dynamic OD matrix and the present condition of the research on traffic flow distribution theories, puts forward algorithms of the dynamic OD matrix estimation based on turning flow model, mainly researches on turning flow model of various type of intersections and dynamic traffic flow distribution model. The main work has been done as follow:At first, it reviews the researches on the dynamic OD matrix estimation model at home and abroad, analyses concretely on the key techniques of the dynamic OD matrix estimation such as reducing the dimensions of feasible solution space, choosing the dynamic traffic flow distribution models and analyzing the traffic flow distribution matrix, puts forward the frame of the dynamic OD matrix estimation based on turning flow model.Secondly, it introduces several methods of getting turning flow at intersections, according to the structure characteristics and traffic flow characteristics of typical intersections, sets up the turning flow model respectively: for eight mutually signal intersections, it is established through the synchronous processing of traffic lights model and detector model; for four mutually and two mutually signal intersections, it is set up by combining computer simulation and mathematics; for no signal intersections, it is built up as a multiple objectives programming model based on genetic algorithm.Then, it introduces the calculating method of travel time and builds up the postpone model, puts forward the traffic flow distribution model based on turning flow through analyzing the road resistance function, the kth shortest path algorithm and the present research condition of traffic flow distribution algorithms, establishes the adjust model to get the selection rate based on the overview of methods to gain the selection rate.Finally, it gives the algorithms of the dynamic OD matrix estimation based on turning flow model, and introduces the related technique problems of the procedure realization such as the network description and kth shortest path algorithm.