考虑电梯排队系统影响的多层流通型配送中心货位指派优化

发布时间：2018-05-08 13:08

本文选题：PH排队模型 + 多层配送中心　；参考：《西南交通大学》2017年硕士论文

【摘要】：多层配送中心并不是单层配送中心的简单叠加,由于受到垂直方向产品运输的影响,其分拣作业比单层配送中心更加复杂。因此,本文在充分考虑下行电梯排队系统影响的基础上,基于分阶段建模的思想,建立了以分拣效率为主要目标的多层配送中心货位指派优化模型。由于电梯排队系统的相关研究存在"强假设不符合实际,弱假设接近实际却难以推导性能指标"的问题,而多层货位指派问题则是缺乏对电梯排队系统影响分拣效率的量化分析。因此,本文首先采用具有良好通用性和解析性的PH分布,将下行电梯描述为一系列相关的PH(ζ)/PH(0,δ]/1/C,C= δ + C1批量到达、批量服务排队模型,并将其等效转化为到达间隔分布与系统状态x相关的PH(x)/PH/1/C排队模型,进而利用QBD过程理论以及矩阵几何方法求得电梯排队系统的平均排队长度、平均等待时间等指标,并分析相关参数对系统指标的影响。其次,考虑电梯平均等待时间的影响,建立了分阶段的多层货位指派模型:先根据产品间的相关性进行聚类,"产品群内相关性最大"为目标将产品划分成与楼层数量相同的产品群;然后,分别以"货位重心最低"和"总分拣时间最小"为目标构建了两阶段模型,将产品群指派到各楼层,再对其层内货位进行优化,分别设计遗传算法和改进型混合差分进化算法对模型进行求解并分析模型参数的影响。最后,基于多层配送中心A的案例数据,运用上述模型和算法进行了货位指派优化,验证了上述电梯排队模型和货位指派模型及算法的有效性。通过模型分析,配送中心在货位指派时应考虑多种因素,尽量将质量较大的产品存储于较低楼层或较低货位以降低产品重心;尽量将周转率较大的产品存储于较低楼层或距离出口较近的货位以减少搬运距离,从而提高作业效率。此外,确定电梯数量和载重时,应考虑总分拣时间最小,而不是某层楼的电梯等待时间最小。
[Abstract]:Multi-layer distribution center is not a simple superposition of single-layer distribution center, because of the influence of vertical product transportation, its sorting operation is more complex than single-layer distribution center. Therefore, on the basis of considering the influence of the downlink elevator queuing system, based on the idea of phased modeling, this paper establishes a multi-layer distribution center location assignment optimization model with the main objective of sorting efficiency. Because of the problem of "strong hypothesis does not accord with reality, weak hypothesis is close to reality but it is difficult to deduce performance index" in the research of elevator queuing system, However, the problem of multi-layer cargo location assignment is a lack of quantitative analysis of the elevator queuing system affecting the sorting efficiency. Therefore, in this paper, the PH distribution with good generality and analytical property is adopted first, and the downlink elevator is described as a series of related PH (味 / P -PH0, 未) / 1 / C ~ (1) C ~ (1) / 未 C ~ 1 batch arrival, batch service queuing model. It is transformed into the PH(x)/PH/1/C queuing model which is related to the arrival interval distribution and the system state x. The average queue length and the average waiting time of the elevator queueing system are obtained by using the QBD process theory and the matrix geometry method. The influence of related parameters on the system index is analyzed. Secondly, considering the impact of the average waiting time of the elevator, In this paper, a multi-layer location assignment model is established: firstly, the product is clustered according to the correlation between products, and the product is divided into the same product group with the same number of floors as the goal of "the greatest correlation in the product group". Taking "the lowest center of gravity" and "the minimum total sorting time" as the targets, the two-stage model is constructed, which assigns the product group to each floor, and then optimizes the cargo position in the layer. Genetic algorithm and improved hybrid differential evolution algorithm are designed to solve the model and analyze the influence of model parameters. Finally, based on the case data of multi-layer distribution center A, the above model and algorithm are used to optimize the cargo location assignment, and the validity of the above elevator queuing model and cargo location assignment model and algorithm is verified. Through the model analysis, the distribution center should consider a variety of factors in the allocation of goods, and store the products of higher quality in the lower floor or lower level as far as possible in order to reduce the center of gravity of the products. Try to store the products with higher turnover in lower floors or near the outlet to reduce the handling distance and improve the efficiency of the operation. In addition, when determining the number and load of elevators, the minimum total sorting time should be considered, not the minimum waiting time for elevators on a certain floor.
【学位授予单位】：西南交通大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：F252.1

【参考文献】