The Study Of The Train Optimization Scheduling Problem In A Railway Depot

In the production process of the refinery, combination and scheduling of the oil tank carriages is time and resources-consuming. At the same time, it demands high level of safety in performance. The optimization of combination can cut down the cost and meet the demand of the safety. The result of the combination of the oil tank carriages has direct influence on the performance of the refinery and its benefits. At present, the work of combination is more relied on the dispatcher's experience, which usually causes low efficiency, danger and other problems So, the combination and scheduling of oil tank carriages, regarded as the key point on the road to the logistic optimization and the management of supply chain, is a very important project for refinery industry.In this paper, by stating the features of the problem of combination of oil tank carriages and researches in China and abroad, a basic model to solve this type of problem is proposed, The coupled problem of the oil tank carriages in scheduling is decoupled by a method by which the latest times of the carriage being moved will be calculated after one carriage's move. According to the specificity of this problem, an improvement is made on the code level of the genetic algorithm, which solves the problem reasonably in relatively short time. The cross and variance operation guarantees this algorithm feasible in practically. In simulation, the influence on the final result caused by the change of the main parameters in genetic algorithm is discussed. The shortage of this algorithm when processing large-scale problem is also analyzed and the new improvement is raised. After the comparison between different sets results, the improved algorithm is verified to be better on performance.Based on the fundamental model, several complex problems of combination of oil tank carriages are explored. These problems are converted into fundamental model and then solved. In those researches, a new element is added which is used to record the tracks of locomotives. By doing this, the algorithm can also work even though the times of the carriages being moved are same. When there is only one direction for the trains'in-and-out, all of them are mixed together as one train and then the problem can be solved as the basic model. In the case of two directions for the trains, the random method is used to obtain the location of the carriage in the first step. Then, this problem is converted into the one-direction problem mentioned before.In the end, the work in this paper is summarized and the further step of this research is clarified.