Research on railway scheduled automobile trains formation plan

With the expansion of the scale of railway automobile transport, it will gradually establish a perfect railway automobile logistics network system throughout the country and the formation plan is more and more important. Timeliness of transport is the key of railway market competition, and operating scheduled automobile train is an important method to improve the timeliness. Fixed vehicle bottom cycle is the best method to operate scheduled automobile train on the railway lines which have balanced wagon flow on the both direction. But it couldn’t realize fixed vehicle bottom cycle on the lines which don’t have balanced wagon flow on bothway. And only organize the through transport of loaded trains has not meet the demand of transportation. Organization of empty wagon flow is as important as the loaded. Therefore, we should combine organization of loaded wagon flow with the empty wagon flow to realize the maximum profit of railway transportation cooperate on the base of meeting transportation demand of customer. The paper analyzed the influencing factors of formation plan based on the theory of wagon flow organization and developed the model of formation plan for empty and loaded wagon. The principle is forming a train when there is enough wagon flow. It takes for example, the transportation organization between central stations and handling stations on a railway line and gets the formation plan and empty formation plan by combining with the model of formation plan for loaded wagon and the model of formation plan for empty and loaded wagon. The aim is achieving the most benefit on the base of satisfying the condition of scheduled train. The optimization model of formation for loaded wagon and the comprehensive optimization model of formation plan for empty and loaded wagon are both zero-one quadratic programming models. There is no best way to solve the zero-one quadratic programming so far. A method to solve nonintersection constraint zero-one quadratic programming model with genetic algorithm is established. The detailed procedure and numerical experiment of the GA are described. The optimal solutions could be obtained easily and fast. It is also suitable for solving large scale problem. A wide application of the method is expected.