The Study Of Optimized Models And Algorithms Of Dynamic Fleet Scheduling Problem

The road freight enterprises face the severe questions of deficient freight sources, the higher empty-loading ratio of vehicles. How to improve their managing level, to organize and schedule vehicle effectively, to low the running cost of the enterprises and strengthen core advantages has been an urge problem to governors of road transport. And the fleet scheduling management is a very important factor which can impact the quality of operating management of road freight enterprises, which also is the studying emphasis of the paper.Firstly, the paper makes use of Multi-Phased Decision and Dynamic Programming theory systemically analyzes the scheduling forms of fleet and the factors which affect the scheduling, and also analyzes the whole scheduling process at the basement of at the basement of analyzing the studying production of the dynamic fleet scheduling both here and abroad. After these works done, the whole serving cycle is divided to many periods, and the paper designs relevant dynamic scheduling models aim to the fleets which own single vehicle-type or multiple types in the sense of making optimize the mission plan as a breakthrough.Secondly, the paper also designs corresponding genetic algorithm to get the results. At the same time, the paper utilizes the Mock Distributed-Genetic Algorithms which design relevant dividing-strategy of the sub-colony, the commutative information model and the commutative information frequency in order to improve the efficiency and evolving quality of the algorithm. The simulated experiment indicates that the calculative speed of the Mock Distributed-Genetic Algorithm is a/one half of the standard Genetic Algorithm. And the quality of results also is superior to the standard one.Successively, the paper farther brings forward the rolling horizon scheduling algorithm for the dynamic fleet scheduling management, and analyzes the main factors which affect the rolling horizon scheduling and designs the relevant the relevant rolling horizon procedure and the adjustive strategy after dealing with the fleet scheduling management at the single-period, for dealing with the missions which appear in the whole periods at random.At last, the paper develops a simulated platform for the rolling horizon scheduling of the dynamic fleet management with the computer technologies of VB and c++ languages. And the paper also does the experiment of testing the performances both the standard genetic algorithm and the mock-distributed genetic algorithm and the experiment of testing the influence of the scheduling at the conditions of different lengths of the mission plan and the experiment of testing the empty-load ratio of the fleet at the whole periods with using the rolling horizon scheduling. The results of experiments prove that the rolling horizon scheduling algorithm in this paper reduce more 5.26% at the empty-loading ratio of vehicle than the social average empty-loading ratio. So the results can prove that the algorithms in the paper is effective and practicable, owning the important value for actual applications.