电子商务环境下并行分区拣选系统的订单合并优化研究

发布时间：2018-05-16 10:35

本文选题：电子商务 + 分区拣选系统　；参考：《大连理工大学》2014年硕士论文

【摘要】：传统仓储操作中,人工订单拣选系统成本占仓储总成本的55%以上,而电子商务环境下,日订单量可达十几万单,且订购的商品种类繁多。为提高订单拣选系统效率,从拣选环节加快订单配送速度,本文针对电子商务环境下的订单拣选系统展开研究。考虑到订单合并策略和分区拣选策略对提高拣选系统效率有效性更高,本文主要研究并行分区拣选系统的订单合并优化问题。当前订单合并策略可分为以下五种：优先规则算法、种子算法、节约算法、数据挖掘算法和启发式算法。由于传统种子订单没有从全局考虑订单间的相似性,本文首先分析了基于相似度聚类的订单合并问题,借用种子算法中合并订单规则其中的一个系数——最大相同通道数,构建基于相似度的订单合并模型,从全局最优的角度依据相似度系数进行订单合并规则构建,最终求得订单合并批次结果；针对并行分区的订单合并问题,研究主要针对分区拣选的数量、大小、存储商品数等因素对拣选系统效率的影响。虽有学者提出工作量均衡对分区拣选策略的重要性,但当前多采用商品合理分配的方法在规划期解决该问题。本文从实际操作角度,构建最小化整体的工作量和平衡各个分区的工作量的订单合并模型,并采用双目标遗传算法进行模型的求解。本论文的主要研究内容如下： (1)从电子商务物流出发,通过对运作流程的梳理,说明订单拣选作业在电子商务物流中的重要作用；同时总结了订单拣选中可能存在的优化问题,包括分拣系统的设计、仓库布局的优化、分拣作业方式和拣货路径策略；最后归纳了当前较常见的几种订单拣选路径策略。 (2)对传统单区型布局拣选系统的订单合并策略进行了定义,并介绍了以最小化总体行走距离的通用订单合并模型；将聚类的思想引入到订单合并问题中,借鉴种子算法的最大相同通道系数为聚类相似性系数,构建了基于相似度聚类的订单合并优化模型；采用改进种子算法构建基于相似度的订单聚类规则,从全局的角度搜索相似度较高的订单,每次进行订单合并时,都选取订单集合中相似度最高的两个订单；最后在五种不同的订单环境下,基于四种评价指标,将聚类规则与种子算法、先来先服务算法进行比较分析,说明了模型和算法的有效性。 (3)在对单区型布局拣选系统研究基础上,对并行分区拣选系统的特征及运作流程进行分析,说明工作量均衡在并行分区拣选系统运作中的重要性；分析了订单完成总时间和各分区工作时间的特点,构建并行分区拣选时间序列模型；考虑到分区工作量均衡对拣选系统的重要性,构建最小化订单完成总时间和各分区工作时间标准差的双目标订单合并模型；采用双目标遗传算法求解此模型,通过基于排序的表现矩阵计算染色体的适应度,并依据染色体的适应度大小和种群整体性能的比较确定其交叉和变异的概率,保持种群的多样性,并控制种群向优解的方向进化。最后在三种不同的订单环境下,将结果与不考虑分区工作量均衡的单目标遗传算法做了比较分析,说明模型在平衡分区工作量上的有效性。本文针对电子商务环境下的并行分区拣选系统的订单合并问题,分别考虑将相似性聚类和分区工作量均衡等因素引入订单合并模型中,提高电子商务订单拣选效率,有利用提高员工的工作积极性和公平感。研究可以为仓储管理部门提高订单处理速度提供决策支持,为仓储配送中心更人性化的管理方式提供指导思路。
[Abstract]:In the traditional warehousing operation, the cost of the artificial order sorting system accounts for more than 55% of the total storage cost. In the e-business environment, the daily order amount can reach tens of thousands of orders, and there are many kinds of goods ordered. In order to improve the efficiency of order sorting system and speed up the order distribution speed from the picking link, this paper aims at the order sorting system under the e-business environment. Considering that order merger strategy and partition selection strategy are more efficient to improve the efficiency of sorting system, this paper mainly studies the problem of order merger optimization in parallel sorting system. The current order merger strategy can be divided into five types: priority rule algorithm, seed calculation method, saving algorithm, data mining algorithm and heuristic method. Because the traditional seed orders do not consider the similarity between orders globally, this paper first analyzes the problem of order merger based on similarity clustering, and constructs an order Merger Model Based on the similarity degree based on the coefficient of the largest same channel, which is based on the global optimum. The similarity coefficient is built for order merger rules, and the results of order merger are finally obtained. In order to solve the problem of order merger in parallel partitions, the influence of the quantity, size, and the number of storage goods on the efficiency of the sorting system is mainly studied. Although some scholars have proposed the importance of the workload balance to the sorting strategy, At present, the method of reasonable distribution of goods is used to solve the problem in the planning period. From the practical point of view, this paper constructs an order merger model which minimizes the overall workload and balances the workload of each partition, and uses a dual objective genetic algorithm to solve the model.
The main contents of this paper are as follows:
(1) from the e-commerce logistics, through the combing of the operation process, the important role of order picking operation in e-business logistics is explained. At the same time, the optimization problems that may exist in order picking are summarized, including the design of the sorting system, the optimization of the warehouse layout, the sorting operation mode and the picking route strategy. Several more common order picking path strategies.
(2) the order merger strategy of the traditional single area layout sorting system is defined, and a general order merger model is introduced to minimize the overall walking distance. The idea of clustering is introduced to the order merger problem, and the similarity coefficient based on the maximum same channel coefficient of the seed algorithm is used for reference, and the similarity clustering is constructed. The order merger optimization model is used to construct the order clustering rule based on the similarity degree, and the similarity degree of order is searched from the global point of view. Two orders with the highest similarity in the order set are selected each time the order merger is merged. Finally, the four evaluation indexes will be based on the five different order environment. The clustering rule and seed algorithm are compared with the first come first service algorithm, which shows the effectiveness of the model and algorithm.
(3) on the basis of the study of the single area layout selection system, the characteristics and operation process of the parallel zoning sorting system are analyzed, and the importance of the workload balance in the operation of the parallel zoning sorting system is illustrated. The characteristics of the total time and the working time of each partition are analyzed, and the parallel sorting time sequence model is constructed. Considering the importance of the partitioned workload balance to the selection system, a two objective order merger model is constructed to minimize the total time and the standard deviation of the working time of each partition. The dual objective genetic algorithm is used to solve the model, and the fitness of the chromosomes is calculated by the sort based representation matrix, and the fitness of the chromosomes is large. The comparison of the performance of the small and the population determines the probability of its crossover and mutation, maintains the diversity of the population, and controls the evolution of the population in the direction of the optimal solution. In the end, the results are compared with the single objective genetic algorithm which does not consider the partition of the work load in the three different order environments, and shows that the model is in the balance partition workload. Validity.
This paper, aiming at the order merger of the parallel partition sorting system in the e-business environment, introduces the similarity clustering and the partition workload balance into the order merger model to improve the efficiency of the order sorting of e-commerce, and improves the employees' working enthusiasm and fairness. The research can be proposed for the warehouse management department. High order processing speed provides decision support, providing guidance for the more humane management mode of warehousing distribution center.

【学位授予单位】：大连理工大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：F713.36;TP18

【参考文献】