Intelligent Optimal Scheduling For Railway Transportation Of Large-scale Non-ferrous Smelting Enterprise And Its Application

Railway transportation is the main artery of large-scale non-ferrous metal smelting enterprises in our country. It plays a very important role in the logistics chain, as it carries out the key task of transporting materials and guaranteeing regular production. With the development of national economy, the rapid increase in the production scale of non-ferrous metal smelting enterprises, the amount of material transportation is also rapidly increasing. However, the railways related to non-ferrous metal smelting enterprises in our country are quite in small scales. The distribution of railway lines and the mode of dispatching are closely related to the production mode of an enterprise. It adds to the difficulty and the complexity of the enterprise railway transportation to have the problems of dispersed distribution of railway stations, small-scaled dispatching mode, short lines of railways, mixed marshalling of self-owned and rented carriages and the fines caused by the cargo retention. At present, the railways owned by non-ferrous metal smelting enterprises in our country are mainly manually scheduled, making freight stations running at full capacity most of the time, which easily leads to the problems such as low efficiency of train operation, not timely arrival of cargoes or even traffic jams and accidents, and seriously hampers the development of non-ferrous smelting business. Therefore, it is of great significance to do some research on the mode of marshaling and dispatching of cargo trains of large non-ferrous smelting business, based on its characteristics, in order to shorten the turnaround time of vehicles, avoid fines caused by cargo detention and improve the efficiency of our non-ferrous smelting enterprises organizing operation.The thesis sets up a scheduling model of railway transportation based on the analysis of the characteristics of large enterprises rail transportation operation, according to the operation of train sorting and classifying, pickup and delivery, and the demand of dispatching organization of railway transportation. The thesis also studies railway scheduling optimization based on cultural evolution algorithm. Thus the thesis proposes railway shunting planning optimization methods, and placing-in and taking-out vehicle optimization of different layout of the enterprise railways, and applies them successfully to enterprise rail transportation intelligent scheduling systems. The main research and innovative achievements of the thesis are as follow:(1) According to the complex mechanism of business scheduling process for railway transportation network, changing circumstances and the impact of many factors, the characteristics of railway transport scheduling process, transportation scheduling processes are analyzed and studied in order to break up the complex railway scheduling model into sorting and classifying model, tree branch lines model and comprehensive layout train placing-in and taking-out model to reduce the complexity of scheduling model, lay a solid foundation for the model optimization solution. In addition, the actual situation of rail transport lines to improve enterprise rail transport scheduling model universality has been fully taken into consideration in the comprehensive layout train placing-in and taking-out model.(2) Ant colony interactive optimization algorithm is put forward to meet the demand of timely solution to the scheduling model of large enterprise railway transportation, making up for the limitations of classical genetic algorithm. The ant colony algorithm and genetic algorithm are blended into the cultural framework, form the main group space and belief space based on ant colony groups. In the evolution process, the main groups periodically organize the worst individuals to learn the best mode population provided by the belief space, taking full advantage of the excellent characteristics of the information contained in the individual, avoiding the population singularity problem of ant colony algorithm, thus greatly improving the convergence speed.(3) The complex train scheduling process is divided into four sub processes:train formation optimization, sorting and marshalling operation optimization, train route arrangement and operation of fetching and delivering carriages, applying the optimization method to several different sub problems in the train scheduling operation to avoid the complexity of optimization problems and greatly improve the efficiency of solution with the optimization model, considering such problems as low accuracy, heavy burden and poor continuity in manual operation of making railway scheduling plan and the characteristics of backward and forward interrelationship and inter influence of the operation subsystems of large enterprise railway transportation. In addition, it can optimize the process of distribution of materials of the metallurgical enterprises within a certain period of time and shorten the period of time when trains are kept in the train station to apply the hybrid elite ant colony algorithm to coordinate and optimize models. In order to solve the problem of train delay caused by renting the national railway system, the penalty factor is included in the optimization as constraint condition. An adaptive penalty function is designed and combined with the genetic elite ant colony algorithm, which solves the problem of train scheduling with the restriction of time penalty and effectively avoids penalty caused by rented train of enterprise railway transportation.(4) A new blending method is applied, which is called a genetic ant colony algorithm, to solve the problem of transformation of the branch-type pickup and delivery vehicles problem into traveling salesman problem (TSP), according to the characteristics of branch-type-specific line in the company marshalling yard, using graph theory knowledge to abstract the distribution for loading and unloading cargo in marshalling yard into the map of Hamilton. This mode genetic algorithm initially searches the information and generates the initial pheromone distribution, enhances the positive and negative feedback mechanism of ant colony algorithm, and greatly reduces the degree of parameter adjustment of ant colony algorithm. In addition, the ant colony algorithm α, β, p parameters reduce sensitivity to changes of the pickup and delivery vehicles in problem size and improve the robustness of the algorithm after the combination of genetic algorithms and ant colony algorithm, increasing the speed of the convergence of the algorithm. It can avoid falling into local optimization problem to a certain extent to apply the maximum-the smallest ant system (MMAS), during the period of having the ant colony algorithm and to use the local pheromone update and global update rules at the same time. (5) Rail transport intelligent scheduling system for large non-ferrous smelting enterprises is designed and developed, which ensures machine drawing in the process of enterprise rail scheduling, on line optimization of dispatching operation and the off-line simulation function. The system increases shunting efficiency, meanwhile, accelerates vehicle turnover effectively and reduces transportation costs of enterprises with the conditions of ensuring the safety and meeting production demand.