Research On Optimization Of Freight Transport Service Network Design Under Integrated Transport System

This paper focuses on the combination of theory and practice, and makes use of optimization theory and methods as tools to investigate the problems related to how to build a freight transport service network under integrated transport system. This paper primarily concentrates on freight transport service network design problem under integrated transport system; the research content covers the area of service network weight handling, path search algorithm research, traffic distribution, frequency and dynamic service network design models and algorithms. Finally, the research result is verified on actual freight transport service network operated by an express company.As to the presentation of freight transport service network under integrated transport system, this paper uses methods of T label and L label to deal with the network weight, combining the "step preprocess" strategy, an improved path search mechanism is then proposed to get the shortest path in network. Furthermore, the deviation search idea is used to design a new K-shortest path search algorithm which can avoid the circuit. After that, a traffic distribution model which considers the optimization of transport cost and transfer cost between different services is built, the model also considers some key constraints. This paper then raises a new sovling mechanism to construct the Lingo script dynamically; this mechanism can solve the traditional problems caused by workload and logic complexity. It can not only ensure build the corresponding script efficiently and accuratly, but also make full use of sophisticated optimization platform and technology.As to the frequency service network design under integrated transport system, the framework of service frame graph is proposed, and then the method is given to determine the transport path set and the design idea of services set is analyzed. Accroding to the types of decision variables, when it is derived output type, the design model based on the service frame graph is built, which belongs to multiple commodity network flow model; when it is non-negative integer type, the design model based on services selected strategy is built. Considering the services choose optimization and transfer amount of freight, the service selected variables which adopts the K-shortest "node" path idea is designed. The work can solve a series of optimization problems related to the freight tranposrt service netwok design under integrated transport system. In views of solving algorithm, a dynamic Lingo script building mechanism and intelligent algorithm mechanism which uses adaptive immune clonal selection algorithm. The intelligent algorithm uses feedback strategy to improve the clonal selection algorithm are designed. It works properly because of the characteristics of service network design; the feedback mechanism has not only improved the sovling speed of the algorithm, but also has ensured the optimization direction of the algorithm.As to dynamic service network design under integrated transport system, taking into account of the lack of traditional time-space network which cannot consider the "freight assembly" strategy in freight transport area, and based on the relationship between service line and marshalling unit, this paper proposes three layers of time-space network, the combination of service layer, marshalling unit layer and freight traffic layer which can solve the problems of service selected optimization, marshalling strategy optimization and freight traffic distribution optimaztion integrately. This paper analyzes the transfer scenario in three layers of time-space network, constructs transfer arcs which will be used to optimize the transfer cost and builds the corresponding mathematical model. The objective function not only considers the optimization of service design cost and marshelling unit constructed cost, but also considers the optimization of transport cost and transfer cost. For the solving algorithm, adaptive immune clonal selection algorithm is used to solve the optimization problem of service design solution. In each iterative step, at first, the marshelling unit constructed cost is added to marshelling arcs, two-stage algorithm is designed to solve the traffic distribution problem between freight traffic layer and marshalling unit layer:at first stage, the column generation algorithm is used to solve the traffic distribution, at second stage, a neighborhood search algorithm based on a closed loop mechanism is used to get better neighbors. After two-stage algorithm solving, the service design solution is updated, then the adaptive immune clonal selection algorithm is used to continue optimizing the service design solution until meet the end condition. At last, the solving algorithm outputs the optimal solution as optimal service design solution.