Research On Wagon-flow Allocation Problem At Railway Marshalling Station In Dynamic And Uncertain Environments

The support system on the railroad yards dispatching decision mainly solves automatic planning problem of daily shift working plan,stage plan and shunting operation plan,at the same time,the level of intelligence plays a decisive role on improving railway freight transportation ability.Among them,the formation of wagon-flow allocation plan which includes of stage plan is the key problem to coordinate the operation of the whole scheduling plan system.All the time,it is the focus for experts at home and abroad that searching for modeling method and optimization algorithm meeting the demand of above problems.The existing research mainly adopted two kinds of methods for modeling:one is building nonlinear mixed integer programming model for the whole problem on the basis of overall consideration of the allocation of wagon flows,the sequence of cars?sorting and assembling,the arrangement of dispatching engines and the utilization of arrival and departure tracks;the other is to decompose the problem into several sub-problems and modeling respectively.But in the process of solving,considering the NP-hard or NP-completeness in this problem,major researches relax and simplify the model,and design a variety of heuristic algorithm for approximate solution.As the problem of forming wagon-flow allocation plan essentially belongs to robust real time scheduling with large,complex and uncertainty factors,the existing models and algorithms are not deep enough in the aspects of the assignment methods proccessing of dynamic,real-time and efficient simultaneously,the theory proving of priority scheduling rules and the analyzing of uncertainty factors.However,besides the complexity of the problem itself,different countries,regions and operation flows make it differ on description methods.It also brings a lot of difficulties to the modeling,analysis and solution of the problem,and restricts the research and development of marshalling station scheduling problem.Therefore,according to the forming characteristics of wagon flow allocation and the optimization theory of production scheduling,related research mainly includes the following five aspects:1.The modeling method of the wagon flow allocation problem based on the production scheduling theory was proposed.In order to build a set of unified modeling method,this dissertation unified the description method of operation procedure,variable naming,problem description,objective function and constraint condition,and a standard construction method of one switch engine,one hump model was given according to the production scheduling theory.2.The multi-stage decision model of dynamic wagon-flow allocating was built based on the method of Greedy.For finding the algorithm of wagon-flow plan in classification yards with the characters of strong searching ability,high solving efficiency,and timely adjustment,by grouping orders,this dissertation raised division method of dynamic sorting interval based on?t_i of each stage and makes more sorting trains might get ahead of break-up shunting locomotive resources in stages.In each sorting interval,optimal selection algorithm of sorting trains was designed which based on several different rules and strategies by making train demand as the matching targets.In order to verify the validity of the algorithm,contrast tests were performed by choosing different strategies or changing the parameters.3.Aiming at the multi-stages heuristic algorithms on basic wagon-flow allocation model,the dissertation provided the equivalent or approximate transformating relations between the optimal goal and the various scheduling strategies by the methods of mathematical proofs.Firstly,it made the optimal goal to equal to the goal of all the departure trains getting the maximum number of cars in the stage plan time,and on this basis,it gave the corresponding dispatching rules from two aspects,which were the selection of sorting sequences and the allocation of the wagon-flows.Then,it presented a strategy that cars having the directions which meet the demands of multiple departure trains could be priority assignmented to departure trains in time order,and then it provided the scheduling rules for making rolling plan.At last,it put forward a nature which could guarantee the stage optimal solution being the globally optimal solution by the directional characteristic of arriving and departing trains.According to these scheduling rules,a heuristic algorithm was designed to obtain the maximum number of cars in each train,and the effectiveness of the rules was verified by an example.4.To get a reasonable sequence of the different values,the similarity measures of vague sets(values)can be carried out using the score function which contains the unknown and uncertainty informations.The dissertation exposed the ranking limitation of score function by exploring the connotation of score function of vague sets and appraising the strength and weakness of various existing score functions.Through an exhaustive enquiry into the connotation of the undefined and unascertained information of vague value as well as an expected definition of vague value,a new category of score function was to be contrived which correlates various vogue value with a certain value of fuzzy sets.When there is an equal score function value,a second ranking is to be executed with definite information.The matching degree between the function and the real world cognition and the uniqueness of the ranking were verified by the data of the instance.5.The dissertation used multi-criteria decision theory based on vague sets to solve the problem of sorting train selection problem at the current stage of the disintegration intervals.Firstly,from the point of systems theory,it builded the hierarchical index system which considers the factors include of the staffs,the engines,the humps,the stracks,the arriving and departing trains and the weather etc;secondly,it presented operation relations of job time?s additive and multiplicative,at the same time that provided the membership functions and non-membership functions which could transform each index datas into vague values;thirdly,according to the logical relation and priority grade of different targets,it designed many kinds of evaluation functions based on the rules,the vague synthesis relations and the weights;fourthly,it adopted the scoring function to convert the evaluation function of each sorting trains,and select the arriving train that has the maximum value calculated by the score function to disintegrate;finally,the effectiveness of the method was verified by an example.This dissertation mainly studied on the method of standard model building on the scheduling problem of wagon-flow allocation under the uncertain condition,the method of the multi-stage wagon-flow allocation scheduling without backtracking based on Greedy algorithm,the dynamic sorting sequence generating algorithm,the theoretical proof of heuristic priority dispatching rules,the design of vague sets?score functions,and the selecting algorithm of sorting trains based on the optimality theory of vague sets.The feasibility of the models and algorithms was demonstrated by experiments or examples,and the effectiveness of the rules was certified by theoretical proof.The research provided a comprehensive solution for searching the optimization wagon-flow scheme which met the actual demand.