Optimization Methods And Applications Of Multilevel Multimodal Transportation Hub Location

The implementation and promotion of “the strategy of transportation power” requires the construction of multi-level integrated transportation hub system aiming to realize zerodistance passenger transfer and seamless freight connection.As the carrier of passenger and freight transfer in the transportation system,the location decision of transportation hubs directly affects the operation efficiency of the whole transportation system.Hierarchy and multi-mode are the basic characteristics of the transportation system in China.The research of multilevel multimodal transportation hub location(MMTHL)conforms to the needs of contemporary transportation development,and can provide theoretical support and decision-making basis for promoting the construction of integrated transportation hub and transportation power.Most of the existing researches on the location of transportation hub focus on the single-layer and single-mode based hub network topology,and ignore the influence of parameter information on the location decision of transportation hubs.Therefore,this thesis studies the problem of MMTHL systematically from different dimensions of parameter information.Based on the traditional hub location theory and method,this thesis studies the optimization methods and applications of MMTHL.Firstly,the location problem of comprehensive transportation hubs are modeled from the theoretical level.The elements of multilevel hubs and multimodal transportation in the transportation hub system are further considered,and the programming methods based on parameter information are applied to analyze the location scheme of comprehensive transportation hubs.Then,efficient algorithms are designed respectively,according to the characteristics of the proposed models.Lastly,from the application level of passenger and cargo transportation,the proposed optimization methods of MMTHL are applied to the urban and rural public transportation system and cargo transportation system.The main works of this thesis are as follows:(1)The bi-objective optimization problem of MMTHL based on complete parameter information.Firstly,based on a Complete-Star-Star multilevel multimodal hub network,a bi-objective optimization model with complete parameter information is proposed for the transportation hub location to simultaneously minimize the total cost and the latest arrival time.Secondly,after analyzing the characteristics of the model,the (?)-constraint reconstruction method is used to obtain the Pareto optimal solution sets of small-scale problems.Furthermore,two heuristic algorithms,a bi-objective variable neighborhood search algorithm and an improved non-dominated sorting genetic algorithm,are designed to solve large-scale instances with a short time.Finally,a series of numerical experiments are carried out based on the data of Turkish cargo transportation network,which verified the effectiveness of proposed bi-objective programming model and the efficiency of the designed algorithms.(2)The stochastic optimization problem of MMTHL considering uncertain demand and direct link strategy.Firstly,based on a hybrid Hub-and-Spoke multilevel multimodal hub network,a stochastic programming model considering the uncertainty of transportation demand and the direct links between non-hubs is established for the transportation hub location,aiming to minimize the total cost including construction cost and transportation cost.Secondly,the model is transformed into an equivalent second-order cone formulation under some mild assumptions,utilizing the expectation theory and chance constraint method.Next,a memetic algorithm combining the local exploitation strategy is designed to solve the large-scale instances with short computational times.Finally,the proposed model and the deigned algorithm are applied into the Turkish cargo transportation system,and a series of numerical experiments are carried out,whose computational results presented good performances of the stochastic programming model and the memetic algorithm.(3)The stochastic optimization problem of MMTHL under uncertain time and cost.Firstly,based on a Ring-Star-Star multilevel multimodal hub network,a stochastic programming model considering the uncertainty of transportation time and construction cost is proposed for the transportation hub location.Secondly,under the assumption of normal distribution,the model is transformed into an equivalent linear formulation applying the expectation theory and chance constraint method; for general parameter distributions,the Monte Carlo method is used to simulate the objective expectation and chance constraints.Also,a memetic algorithm combing the shift searching strategy is designed to solve the large-scale instances efficiently,which is suitable for the Monte Carlo simulation method too.Finally,the data of CAB and Turkish transportation network are used to perform numerical experiments,whose computational results validated the proposed stochastic programming model and the designed memetic algorithm.(4)The disributionally robust optimization problem of MMTHL for dual uncertainty and cluster-based strategy.Firstly,based on a Complete-Star-Star multilevel multimodal hub network,the uncertainty characteristics of travel time,construction cost,and the corresponding distribution functions are analyzed,and a disributionally robust optimization model considering the uncertainty of time and cost,and the cluster-based strategy,is established for the transportation hub location.For the convenient of calculation,the bounded perturbations and Gaussian perturbations are constructed,and a safe tractable approximation and an equivalent transformation of the distributionally robust optimization model are obtained,respectively,which can be solved by the CPLEX software only for smallscale instances.To solve the large-scale instances quickly,two heuristic algorithms,a variable neighborhood search algorithm and a population-and-search based heuristic algorithm,are designed.Finally,the models and the algorithms are applied into the urban and rural public transportation system of Guangrao,Shandong Province,and large-scale instances generated randomly,whose computational results verified the superiority of the model and the efficiency of the designed algorithms.