Model And Algorithm Of High-dimensional Multi-objective Discrete Transportation Network Design

With the rapid development of city motorization, the contradiction between traffic demand and traffic supply continues to expand, resulting in many traffic problems such as traffic congestion, traffic pollution and frequent traffic accidents. Transportation planning is a scheme that provides transportation facilities, if it can be made scientifically and reasonably, the occurrence of traffic problems can be avoided or reduced from the source. However, transportation planning is commonly a qualitative method, which lacks considering user’s behavior on the actual network. After implementing this kind of planning scheme, some bad phenomena often occur, such as traffic improvement was not obvious even some traffic problems were exacerbated, moreover, the limited investment funds were wasted. Therefore, under the condition of the limited investment funds, based on considering the users’ behavior on the network, the quantitative research on the transportation network design problem has important application value.Firstly, the thesis deeply analyzed the related basic theory of high-dimensional multi-objective transportation network design problems. And then, based on expatiating bi-level programming model, multi-objective optimization model and typical transportation network design model, from the perspective of many departments of government decision-making and users’ travel behavior, a discrete transportation network design bi-level programming model that simultaneously optimizes five objectives of system impedance, exhaust emissions, investment costs, saturation level and users’ travel behavior was constructed. Among which, the upper level programming model was constructed by changing multi-objectives into multiple single objective, and the constraint of link capacity increment was set according to planning capacity. The lower level programming model was constructed using the user equilibrium assignment model under a fixed traffic demand.Then, in order to solve the model and ensure good convergence and distribution of solving algorithm, based on the second generation of non-dominated sorting genetic algorithm (referred to as NSGA-II), the algorithm was improved based on the kth small distance double truncation strategy, in other words, the improved NSGA-II was designed, and its corresponding program was developed using Matlab language.Finally, taking the Nguyen-Dupuis network as an example, through comparative analysis of the results of NSGA-Ⅱ and improved NSGA-Ⅱ on the three-dimensional and four-dimensional multi-objective discrete transportation network design model, it proved that the improved NSGA-Ⅱ can effectively solve basic high-dimensional multi-objective discrete transportation network design model, combined with the relevant professional knowledge,8 eventual traffic network design optimization scheme was got.