| Urban transportation is a crucial factor in urban development and progress. It complements social and economic development. If the development of transportation is ahead of social and economic development, likes in the most of the developed countries, it will lead economy a long-term rapid development. Otherwise, it will become a major bottleneck that restricts economic development. That is the reason why, especially for China, solving transportation problems is very important and cannot be ignored both now and in the future. Survey is the basic and necessary method to prepare the transportation planning. Traffic data obtained from survey effects directly whether a plan is authentic and reliable. Since the amount of traffic source is very huge in most of transportation plans, traffic analysis zone is needed to act as the basic spatial statistical unit. It requires that zone cannot be too small to waste resources and also cannot be too large to ensure the accuracy of transportation plan. Thus a valid and effective traffic analysis zone delineation method is needed.Accordingly, under the background of urban transportation planning, this dissertation proposes a serial of methods from the views of data abstraction methods, traffic analysis zone problem optimization modelling, optimal solving traffic analysis zone problems both with and without zone number decision, by techniques of integer programming. The major works of this dissertation are list as follow.(1) In terms of the empirical traffic analysis zone partition methods in actual transportation planning activities, data abstractions methods, the content and models of optimization objectives, modelling framework and solving methods of Optimal Traffic Analysis Zone Delineation Problem (OTAZDP) in theoretical research of traffic analysis zone delineation problem, a comprehensive literature review is given. Based on that, some extended works on algorithms designing and modelling are described in detail. This part of research involves:First, regarding to the traffic analysis zones in the practical transportation activities, the reason why a traffic analysis zone system should be optimized is discussed based on literatures, and the delineation business rules in practice are summarized and reviewed. Second, from micro-perspectives, regarding the theoretical research of traffic analysis zone delineation problem, for the problems of what optimization objectives should be chosen and how to abstract a traffic analysis zone delineation scenario, more detailed reviews about study area discretization methods, accessibility measurement methods of spatial units, contents and formulations of optimization objectives are presented. According to that, a adjacent matrix construction algorithm based on a proposed z-style coding, local depth calculation algorithm of spatial units and two optimization objectives based on local depth value measured accessibility-a minimize zone weighted travel cost objective of OTAZDP and a minimize zone accessibility heterogeneity objective of OTAZDP- are put forward. Third, from macro-perspectives, regarding the typical optimization framework and solving methods of traffic analysis zone delineation in theoretical research, reviews are described and discussed. The optimization framework is summarized as five elements, including input data, optimized objective, basic constraint, context constraint, algorithm and evaluation indicator. According to that, more detailed reviews about research categories from the perspectives of both input data and algorithm are presented.(2) In terms of how to ensure contiguity constraint of OTAZDP by integer programming technology, a brief literature review is given. According to that, regarding integer programming modelling method of contiguity constraint of OTAZDP and its influences on the computational efficiency of OTAZDP, the works of modelling and case studies are implemented. This part of research involves:First, based on the literature reviews of OTAZDP contiguity constraints modelling methods, three presentation, ordered path presentation and flow presentation based contiguity constraint modelling methods are improved to cope with OTAZDP according former researches. Second, a modelling method to ensure first-order contiguity of OTAZDP using adjacent matrix presentation is put forward. Third, the theoretical amount of both variables and constraints of all four models are derived and given. Based on that, computational complexity of all four models are analysed and discussed. Fourth, two cases of small-sized problem is studied to show the advantage of the proposed model for solving large scale problem.(3) In terms of how to formulate a basic OTAZDP in integer programming way under the scenario that the number of zone is fixed beforehand and how to design the algorithms,0-1 integer programming model based on P-Median Problem (PMP), algorithm designs and case studies are given. This part of research involves:First, based on the classical PMP, assuming zone number is given, a 0-1 integer programming model names as TAZ-PMP is proposed, which possesses a minimize zone attribute heterogeneity objective and basic contiguity constraints. Second, three algorithms, including an implicit enumeration algorithm (TAZ-IE algorithm), a lagrangian and surrogate relaxation based local search algorithm (TAZ-LSLSH algorithm) and a lagrangian and surrogate relaxation based column generation algorithm (TAZ-LSCG algorithm), are designed to solve the proposed TAZ-PMP model. Third, using simulation cases that modified from OR-Library and Pcb3038 benchmark database, efficiency of the three algorithms are analyzed and discussed.(4) In terms of how to formulate a extended OTAZDP in integer programming way when considering zone number as a decision variable and how to design the algorithms, models, algorithm designs and case studies are given. This part of research involves:First, an improved approach, in which the OTAZDP is formulated as a mixed integer programming model, is proposed by treating the zone number as an unknown variable so as to avoid the limitations that such number has to be determined in advance in most OTAZDP and provides the optimal solution including both the number of zones and partitions. An overall geographical error is minimized, and homogeneity is guaranteed by the constraints according to delineation context. On the other hand, contiguity check is executed by a predefined adjacent matrix and assures only first-order spatial adjacent relationship for compactness of zone shape. Second, in terms of solving method, the initial model considers the basic geographical unit as a building block, which is subsequently refined by an influence area construction algorithm so as to reduce the OTAZDP size from basic unit scale to influence area scale. After proposing and proving two properties of the optimal zone amount, two domain reduction methods are given. One narrows the range of zone number decision variable by giving the maximum lower bound. The other method provides a way to enumerate all feasible zone number values according to a kind of homogeneity constraints. A non-solution situation of proposed model is explored and is settled by our proposed agglomerative hierarchical clustering-based heuristic algorithm. Third, based on the above contents, two methods to solve the problem are given and are examined using simulation cases that modified from OR-Library and Pcb3038 benchmark database. The results obtained by our proposed approaches constitute good solutions and are generated in a reasonably low computational cost.(5) Using our proposed data abstraction approaches, integer programming modelling methods and solving algorithms, a practical OTAZDP case from Suzhou industrial park is executed and studied thoroughly. |
PDF Full Text Request