基于时空数据挖掘的生鲜配送中心选址研究

发布时间：2018-03-28 08:35

本文选题：生鲜配送中心　切入点：空间聚类　出处：《河北师范大学》2017年硕士论文

【摘要】：伴随着我国经济和生活水平的不断提高,生鲜产品已经成为人们饮食生活中不可或缺的食品之一,它是人身体中各种维生素和微量元素的重要来源,其生鲜品质问题越来越受到关注。生鲜产品易腐败的特点致使它在运输过程中使用冷藏车进行运输,也就是需要在冷链的条件下进行配送。为了保证客户能够买到新鲜的生鲜产品,大多数企业和生鲜零售网点均自建生鲜产品配送中心进行生鲜产品的统一配送来满足客户的需求,因此,生鲜配送中心的选址的优劣直接关系到企业或者生鲜销售网点的盈利和发展,如何科学合理的选择生鲜配送中心的建设地点是本文研究的重点问题。首先,阐述了生鲜配送中心、生鲜生命周期、数据挖掘、数据清洗和配送中心选址的相关基础理论。对生鲜配送中心的功能和类型做了详细的解释,阐述了生鲜产品生命周期函数的基础内容,介绍了数据清洗原理、数据挖掘的主要方法及一般流程,描述了配送中心的选址原则、影响因素以及步骤等,这些基础研究给生鲜订单数据源的挖掘和生鲜配送中心选址模型的创建提供了理论支撑。其次,论述了数据清洗处理的步骤与方法,对空间聚类分析与时间特征分析的算法进行了探讨。在此基础上,利用非等覆盖半径模型确定配送中心的候选可建区域。再利用规划方法以生鲜配送系统的最小成本目标函数构建选址模型,其中成本包括能耗费用、设施费用、用地费用、运输费用和因新鲜度降低造成的损失等几项成本,并提出了模型求解方法。最后,对实例进行应用分析和求解。该部分首先对石家庄生鲜销售网点提供的O2O(Online To Offline)平台的订单数据进行清洗和挖掘,利用空间聚类分析和时间特征分析两种方法分析客户在不同时间段的需求量变化情况和不同区域客户需求量的特点。通过构建生鲜配送中心的选址优化模型,结合数据挖掘得出的客户需求量结果和实地调研得到的数据,使用非等覆盖模型和最小成本目标函数的编程解算,求解出实例的最终结果,验证了选址模型的可行性。然后对现有生鲜网点进行分析评价,从而给生鲜销售网点提供生鲜进货量的决策支持,避免了配送中心库存出现不合理的情况,从而减少不必要的成本损失。本文研究结果从理论性和实践性两个方面给生鲜网点提供了的策略参考。本文基于O2O大数据定量分析,以时间特征、空间特征和客户需求特征等时空多维特征构建了选址模型,具有一定的创新性。但是,本文还略有不足,生鲜腐败函数和选址模型的解算还需要更进一步地细化研究。
[Abstract]:With the continuous improvement of economy and living standard in our country, fresh products have become one of the indispensable foods in people's diet and life. It is an important source of vitamins and trace elements in human body. The problem of fresh product quality is getting more and more attention. Fresh products are prone to corruption, which causes them to use refrigerated trucks for transportation during transportation. In other words, distribution is required under cold chain conditions. In order to ensure that customers can buy fresh and fresh products, Most enterprises and retail outlets set up their own fresh products distribution center to meet the needs of customers. The location of the fresh distribution center is directly related to the profit and development of the enterprise or the fresh distribution network. How to choose the construction site of the fresh distribution center scientifically and reasonably is the key problem of this paper. The basic theories of fresh distribution center, fresh life cycle, data mining, data cleaning and location of distribution center are expounded. The functions and types of fresh distribution center are explained in detail. This paper expounds the basic contents of the life cycle function of fresh products, introduces the principle of data cleaning, the main methods and general process of data mining, and describes the principle of location of distribution center, the influencing factors and the steps, etc. These basic researches provide theoretical support for the mining of fresh order data sources and the establishment of fresh distribution center location model. Secondly, the steps and methods of data cleaning are discussed. The algorithms of spatial clustering analysis and time feature analysis are discussed. Using the non-equal coverage radius model to determine the candidate buildable area of the distribution center, and then using the planning method to construct the location model with the minimum cost objective function of fresh distribution system, in which the cost includes energy consumption cost, facility cost, land cost, etc. The cost of transportation and the loss caused by the decrease of freshness are discussed, and the method of solving the model is put forward. In this part, the order data of the O2O(Online to offline platform provided by the fresh selling network in Shijiazhuang are cleaned and excavated. Spatial clustering analysis and time feature analysis are used to analyze the change of customer demand in different time periods and the characteristics of customer demand in different regions. Combined with the customer demand result obtained by data mining and the data obtained from field investigation, the final result of the example is solved by using the non-equal coverage model and the programming solution of the minimum cost objective function. The feasibility of the location model is verified. Then the existing fresh outlets are analyzed and evaluated to provide the decision support for raw and fresh sales outlets and avoid the unreasonable situation of distribution center inventory. In order to reduce unnecessary cost loss, the results of this paper provide a strategic reference for fresh sites from both theoretical and practical aspects. This paper is based on the quantitative analysis of O2O big data, with time characteristics, The spatial feature and customer demand feature are used to construct the location model, which is innovative. However, there are some deficiencies in this paper, and the calculation of fresh corruption function and location model need to be further studied.
【学位授予单位】：河北师范大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：F724.6;F252;P208

【相似文献】