Study On Dynamic OD Matrix Estimation Model And Arithmetic

In transportation networks, dynamic OD matrix describes the time-dependent traffic demands, which is an important input of ATIS and ATMS, as well as a basic input of DTA models and many applicable microscopic traffic simulators. Given the fact that dynamic OD matrix is difficult to attain in ITS research, this paper reviews the evolution process of the theory of dynamic OD matrix estimation, analyzes the character and existing demerits of different models, and summarizes several key issues in the research of this field. Based on the previous work, the multi-objective shortest paths, variable inequality models based on ideal links and paths, dynamic origin-destination matrices estimation for the intersections, freeways and general networks, traffic monitor optimal locations of OD estimation, are investigated. All of these laid foundation for further studying dynamic OD estimation theory.In general, there is no absolute shortest path for multi-objective shortest path problems in general. In order to get the efficient paths meeting decision-makers' requirements, the model for multi-objective shortest path problem is constructed. The paper develops a polynomial algorithm of the multiple objectives shortest path by combining the k-shortest path algorithm and fuzzy multiple objective lattice-order decision making. By giving an acceptable upper limit for each single objective, this novel algorithm determines the feasible paths set of each single objective seperately by using k-shortest path algorithm. In this way, the efficient paths simultaneously meeting multi-objective constraints, are acquired. Finally these efficient paths are compared and optimized by using the multi-objective lattice-order decision-making method. Then, the path set which a decision-maker is satisfied with is attained. Furthermore , the paper develops a practical algorithm of acquiring the efficient paths for the bi-objective shortest path problems by combining k-shortest path algorithm with bi-objective decision making. The algorithm is polynomial and can work fast to get all the efficient paths.The reason to cause Braess's Paradox is neglecting dynamic guidance users' diverse individual demands. In view of this, a mathematic model with users' individualized demands is constructed based on the rational alternative paths of multi-objective shortest paths.By fusing k-shortest path algorithm and multi-objective decision theory, etc., an efficient algorithm of multi-objective shortest paths is gained. Through carrying out cluster analysis by introducing similar degrees, rational alternative routes under different situations are obtained. This paper does not just provide an "optimum route" based on a single objective for a driver to adopt, but offers many "rational" candidate routes with different attributes for a driver to choose, which suits the driver's actual demands even more.In a dynamic traffic distribution, a variable inequality model is developed for destination nodes based on the ideal sections or paths which satisfies DUO conditions. Then the equivalence between the two is proved. In order to solve variable inequality problem quickly, Mutative Scale Optimization Algorithm using two kinds of chaos is proposed.Through summarizing assignment principles of traffic monitor of OD matrix estimation, a model of traffic monitor optional assignments is developed on the basis of the multi-objective lattice-order decision-making. Due to the complexity of environment and rational limits of decision-makers, it is hard for decision-makers to make total order structure description. By using the lattice theory, the totally ordering description of scheme optimization is extended to the lattice ordering one .Based on decision-making theory, fuzzy set theory and the other related knowledge, the concepts of fuzzy multi-objective single and multiple level lattice-order decision-making are put forward and the corresponding model are constructed, and two kinds of basic methods for the models are proposed. Finally, the problem of traffic monitor's assignment of OD estimation is analysed.For characteristics of dynamic OD matrix estimation of intersections, a model is set up based on discount coefficients and disperse coefficients, and a chaos optimization algorithm is presented based on the Method of Reduced Gradient. For the characteristics of dynamic OD matrix estimation of freeway, models under two different situations are put forward. A high-speed algorithm based on the Method of Reduced Gradient Projection chaos optimization algorithm is designed. For the characteristics of dynamic OD matrix estimation of a general networks, a bi-objevtive model is set up, the Method of Reduced Gradient Projection chaos optimization algorithm is used to solve this problem.