铁路车站取送车系统优化理论与方法研究

Study on Optimization Theory and Method of the Wagon's Takeout and Delivery Scheduling System in Railway Station

取送车系统是车站作业系统的子系统,其效率的高低直接关系到出发列车正点率、车辆周转时间等运输生产指标.基于系统优化理论与方法、组合数学、集合论以及生产实践经验,分析取送车系统的要素、结构、环境和特征,研究取送车问题的建模方法、模型解的构造方法、寻优算法和提高问题寻优效率的启发式方法,并通过算例对研究成果的适用性进行验证.主要研究内容和结论如下.

The Wagon's takeout and delivery scheduling system in railway freight station (WTDSS) is a subsystem of railway station scheduling system, and its operation efficiency has a close bearing on the on-schedule rate of departure trains, wagon turnaround time and many other transportation production indexes. Based on the system optimization theory and methodology, the combinatorial mathematics theory, the set theory as well as some transportation production experiences, this dissertation analyzes the elements, structure, environment and characteristics of the Wagon's takeout and delivery scheduling system and studies modeling method, the construction method of model solution, optimization algorithm and heuristic methods for improving optimization efficiency of the Wagon's delivery and takeout problem. The applicability of the research results has been validated by example. The main research contents and conclusions are as follows. (1) The operation objects, the loading-unloading locations and their layout forms, all kinds of takeout and delivery operations and their component elements, the operation organization modes are systematically analyzed in details. The WTDSS functional factors, environment and characteristics are comprehensively analyzed. The relevant concepts are clearly defined. (2) The Wagon's delivery and takeout problem (WTDP) whose loading-unloading locations show radial layout (R-WTDP) is studied. Under the assumption that the locomotive haulage capacity and the wagon acceptance capacity of each loading-unloading location are both unlimited, the ideal mathematical models for the non-through and the unfixed form wagon flow are respectively established based on the fact that the sequence, moment and batch number problems of takeout and delivery operations during a period of time are a whole system. According to features of the models, expression method of the solution is designed, and modular implementation method for the models is proposed to construct solution and to calculate the value of solution. The scheduling schemes obtained by the models and their implementation method have various adaptive operation organization modes, and therefore overcome the shortcomings caused by using the fixed organization modes. (3) The WTDP whose loading-unloading locations show tree-shape layout (T-WTDP) is studied. Two mathematical models are respectively built for the non-through and the unfixed form wagon flow in the case of several trains successively arrived in the railway station. The solution expression and the modular implementation method for models are presented according to the characteristics of the models. Through the models and their implementation method, the reasonable operation schemes can be found which not only include the reasonable sequences, moment and batch number of takeout and delivery operations in planning stage but also have the adaptive operation organization modes. (4) Based on the research results about the R-WTDP and the T-WTDP, the generic model (G-WTDP) is built for the sequence, moment and batch number problems of takeout and delivery operations during a period of time, which applies to different loading-unloading location layouts, all kinds of arrival or departure wagon flows. And then the WTDP is clearly defined as a combinatorial problem with two optimization dimensions. Aiming at the high computational cost of solving the WTDP, a set of methods used to generate the complete solution is presented by abstracting and standardizing objective constraints in the WTDP and principles arising from practical experience. Through the proposed methods, the search space of the WTDP can be reduced from 2n-1n! to n!, and therefore improve the efficiency of solving problem. The theoretical basis for the determination procedure of batch scheme is discussed from the set theory point of view. Three benchmark examples which are respectively for R-WTDP, T-WTDP and G-WTDP are designed, and their best known solutions are given. (5) To improve the adaptability of optimization algorithm to different situations, a hybrid algorithm with adaptive parameters is proposed. It treats ant colony optimization system (ACO) and genetic algorithm (GA) as its two parallel strategies, and it uses the cloud association rules, grey prediction and feedback from its iterative process to control adaptively relevant algorithm parameters, which can effectively reduce the reliance on the initial setting of parameters. In view of the WTDP characteristics, the time-expectation heuristic information and a novel mutation operator are presented. The former is used as the visualization factor in state transition probability formula of node selection to improve the algorithm convergence efficiency. The latter is used to transform a kind of infeasible solution which cannot meet the time-domain constraint, which makes the obtained solution be or be more close to feasible one. It reduces obviously the redundant computation of the algorithm and therefore improves the searching efficiency.