自适应遗传算法在越库车辆调度问题中的应用研究

发布时间：2019-06-14 13:36

【摘要】：随着国民经济的快速发展,对社会物流需求量显著增加,推动物流产业保持持续、稳定、快速发展。在物流实际运作过程中,越库车辆调度一直以来都是影响企业的运输效率和物流成本的关键因素。越库是指从入库货车到达配送中心、停靠卸载仓门进行卸货开始,货物进行分配和处理,再到装载仓门进行装载的整个过程的一种组织方式,货物不会在配送中心存储而是直接分配的方式进行。这种组织方式的实施,能够从成本、时间、环节等方面降低和减少,从而大大提升了物流的效益。越库车辆调度问题可描述为在一定约束条件下,如何合理分配车辆与仓门,使得整个运作在成本或时间上达到最优化的问题。它是一个典型的NP难(NP-Hard)问题,也是最难求解的经典组合优化问题之一。在过去的几十年里,研究者们都在不断寻找和尝试新的调度算法,使其提高操作效率、降低操作成本和时间、增加企业的竞争力。遗传算法作为仿生学方法中最重要的算法之一,也是应用最广泛的进化计算方法之一。遗传算法在解决各种非线性优化问题中表现出的自适应性,全局最优性以及隐含的并行性等特点让其在调度的优化研究方面有着不可替代的优势。本文针对多仓门越库车辆调度问题特点,以遗传算法为基础,提出了一种改进的、更加适用于求解多仓门越库车辆调度问题的新算法。由于简单遗传算法在应用中常出现收敛过慢、稳定性差以及早熟现象等问题,而现有的一些改进的自适应遗传算法在求解过程中容易产生局部最优解等缺陷。本文从整个遗传算法流程出发,针对遗传算法前期容易陷入局部最优,中后期进化缓慢等缺陷,从种群多样性,个体最优保存策略以及交叉概率和变异概率几个方面进行改进,根据实际问题改进交叉方式和变异方式,提出了一种能有效解决多仓门越库车辆调度问题的自适应遗传算法。通过实验结果表明,该算法在收敛快速性和稳定性等方面都有明显的改善,达到了预期的效果。最后,本文针对越库车辆调度的模型和改进的算法来开发一个多仓门越库车辆调度系统。
[Abstract]:With the rapid development of national economy, the demand for social logistics has increased significantly, and the logistics industry has been promoted to maintain sustainable, stable and rapid development. In the actual operation process of logistics, vehicle scheduling has always been the key factor affecting the transportation efficiency and logistics cost of enterprises. Crossing the warehouse refers to a way of organizing the whole process of loading and unloading the goods from the loading truck to the distribution center, the goods are distributed and processed, and then to the loading door. The goods will not be stored in the distribution center, but will be distributed directly. The implementation of this organization can reduce and reduce the cost, time, link and so on, thus greatly improving the efficiency of logistics. The problem of vehicle scheduling over warehouse can be described as the problem of how to distribute vehicles and warehouse doors reasonably under certain constraints, so that the whole operation can be optimized in cost or time. It is a typical NP difficult (NP-Hard) problem, and it is also one of the most difficult classical combinatorial optimization problems. In the past few decades, researchers have been constantly looking for and trying new scheduling algorithms to improve operational efficiency, reduce operating costs and time, and increase the competitiveness of enterprises. Genetic algorithm, as one of the most important algorithms in bionic methods, is also one of the most widely used evolutionary computing methods. Genetic algorithm has irreplaceable advantages in scheduling optimization research because of its adaptability, global optimality and implicit parallelism in solving various nonlinear optimization problems. In this paper, according to the characteristics of multi-warehouse gate vehicle scheduling problem, based on genetic algorithm, an improved new algorithm is proposed, which is more suitable for solving multi-warehouse gate vehicle scheduling problem. Because of the problems of slow convergence, poor stability and premature phenomenon in the application of simple genetic algorithm, some existing improved adaptive genetic algorithms are easy to produce local optimal solution and other defects in the process of solving. Based on the whole process of genetic algorithm, aiming at the defects of genetic algorithm, such as easy to fall into local optimization in the early stage and slow evolution in the middle and late stages, this paper improves the population diversity, individual optimal preservation strategy, cross probability and mutation probability. According to the actual problems, an adaptive genetic algorithm is proposed, which can effectively solve the vehicle scheduling problem of multi-warehouse gate crossing. The experimental results show that the convergence rapidity and stability of the algorithm are obviously improved, and the expected results are achieved. Finally, a multi-warehouse gate vehicle scheduling system is developed according to the model and improved algorithm of cross-warehouse vehicle scheduling.
【学位授予单位】：广东工业大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：U492.22;TP18

【相似文献】