【作者】 郑宏珍；

【导师】 战德臣；

【作者基本信息】 哈尔滨工业大学 ， 计算机应用技术， 2008， 博士

【摘要】 随着全球竞争的加剧和科学技术的进步,现代管理思想和手段不断变革和发展,越来越多的企业开始注重供应链分销网络性能优化,以提高客户的满意度,提升企业的核心竞争力,从而保证供应链系统的快速响应变化能力,支持供应链整体绩效的改善和个体收益的“双赢”。针对供应链分销网络的选址、产品优化配置、运作和再优化,以供应链分销网络优化为目标,本文以“供应链分销网络选址问题研究→供应链产品的优化配置问题研究→供应链分销网络流段选址优化问题研究→供应链分销网络再优化问题研究”为研究主线,给出了若干模型与算法以解决上述问题。主要研究工作包括以下几个方面:(1)针对面向客户群的供应链分销网络选址问题,采用数据挖掘中的聚类分析的方法,引入聚类有效性函数和划分模糊度的概念,将划分熵与划分模糊度相结合,给出“和”聚类有效性函数,并定义一种修正的划分模糊度作为聚类有效性函数,通过引入聚类的分类属性的权值参数λ来控制数值属性和分类属性对聚类过程的作用,由于聚类有效性问题又可以转化为最佳类别数k的确定,为此针对客户群中客户的数值数据、类属性数据和混合数据,采用基于修正划分模糊度的参数选择方法,研究对应客户数据类型聚类的最佳选址数量的确定方法。(2)针对供应链的特点,以客户偏好序列数据为切入点,分析消费者对多种评价对象的心理趋向是否存在聚类,建立客户偏好取向与客户特征属性间的关联关系模型,借鉴数据挖掘的符号序列聚类方法,研究符号类型序列数据对应的性质,沿着形式化和实例化两个方向讨论符号序列相似性问题,对偏好符号序列聚类问题的本质进行分析;研究如何应用自组织特征映射作为符号序列的聚类算法,并对聚类模型进行比较。(3)针对面向顾客的服务设施选址问题,研究带双重容量限制和时间价格约束的供应链分销网络流段选址问题,给出启发式算法,并将该问题转化为传统的流段选址问题来解决。针对市场需求细分和顾客流量接受绕行距离需求的供应链分销网络选址问题,研究带路径需求服务半径的供应链分销网络流段选址问题,给出解决问题的贪婪算法、局部搜索算法以及禁忌算法,并进行比较。研究需求不确定情况下的的供应链分销网络设施流段选址分配问题,建立MAXMIN-FIFLP模型,给出启发式算法。(4)针对市场的竞争情况,研究当沿途顾客的消费能力随着存在的经营网点的效用函数有界线性扩张时,供应链分销网络流段选址优化问题。构造竞争条件下消费能力线性扩张的CCLE-IFIFLP模型,给出对应的启发式算法。研究竞争环境下的市场份额最大化的选址问题,通过引入竞争设施聚集引起的需求增长率和距离折扣率来刻画设施的聚集效应,建立该问题的模型,给出求解该问题的分支定界算法和贪婪算法,分析需求增长率和距离折扣率对选址决策的影响。同时针对供应链中客户满意度存在的问题,用顾客的时间的满意度来表示覆盖半径,研究基于时间满意度的最大覆盖的供应链分销网络再优化问题,给出问题的模型和启发式算法。为验证理论研究成果,本文以威海海都食品企业信息化作为典型应用案例,应用本文提出的各种算法和模型,实现供应链分销网络的产品分析、渠道分析和市场分析。给出了应用及结果,验证了本文的理论和方法。更多还原

【Abstract】 With intensifying global competition, progress on science and technology, and evolution of modern management ideology and means, for the purpose of improving customer satisfaction and enhancing their core competitiveness, more and more enterprises are starting to pay attentions on the optimization performance of the Supply Chain Distribution Network to guarantee the rapid response and change of the supply chain system, the support of improvement of overall performance and the capability of supply chain system and the win-win situation of the individual income.In respect of locating the Supply Chain Distribution Networks, optimal configuration of products, operation and further optimization of products, this paper will follow the outline of "research on locating the Supply Chain Distribution Networks, research on optimal configuration of products, research on optimization of selecting location for flow sections of Supply chain retailing network, and research on further optimization of Supply chain retailing network”. Some models and algorithms are established to address the issues mentioned above. The main contribution of this thesis includes the following aspects.1) Aiming at locating the customer- group oriented facility of the Supply Chain Distribution Networks, the cluster analysis methods of data mining are adopted, including introduction of cluster validity function and division of fuzzy degree, combining the partition entropy and partition fuzzy degree, giving“sum”cluster validity function, giving modified partition fuzzy degree as cluster validity function. This paper introduces the weight parametersλfor the assification attributes to control the influence of process of clustering from numerical attribues and classification attributes of the classification. Because the effectiveness of the cluster can be transformed into the best category number k, parameter selection method based on modified Partition fuzzy degree is proposed to study the determination method for optimal site chosen number corresponding to the cluster of customer data sructure to deal with the problem from numerical data, attribute data and mixed data type in customer group.2) Aiming at the characteristics of supply-chain, considered the customer preferences sequence data, analysis is carried out to determine whether a cluster exists for customer’s mind tendency for multiple evaluation objects. This paper creates association relation model relating customer preferences and customer characteristic attributes. Borrowed cluster method for the symbol sequence in data mining, corresponding properties of the data from type symbol sequence are studied. The reasearch follows two directions, formal and instantiation, to discuss similarity issues of symbol sequence in which essential problems in cluster from preference symbol sequence clustering are analized. In addition, this paper also studies on how to apply self-organizing feature map as a symbol of the sequence clustering algorithm, and compares clustering model.3) As for the service facility location problem for customers, location selection problem relating retail network flow supply chain with dual capacity constraints and time-price constraints and the heuristic algorithm is given, in which the problem is transformed into a traditional CFIFLP issue to deal with. In view of the market demand has been fine segmented and retail network location selection for the supply chain for the customer flow that accepts detour distance, retail network location selection for the supply chain with route requirement and service radius is atudied, in which a greedy algorithm to solve the problem, local search algorithm, as well as taboo algorithm are given and compared. Research on more uncertain circumstances demand Flow Interception Facility Location Problem, and established a model and MAXMIN-FIFLP model, and gives a heuristic algorithm.4) Aiming at the situation of competition in the market, this paper studies the optimal location selection problem when the consumption capacity of the on road customers expanded. Competitive Capability Line Expand CCLE-IFIFLP model and heuristic algorithm have been proposed. Futther, location selection problem that maximizes the market share under the competitive environment is studied by introducing incremental demand rate and distance discount caused by competitive facilities cluster to describe cluster effects. In this paper, the model has been created, and the branch and bound algorithm and the greedy algorithm are used to solve this problem. The growth rate of demand discount and distance discount rate have been analized to determine the effects to the loation selection decision. To solve the customer satisfaction problems in the supply chain, customers’time satisfaction rate is used as coverage radius to discuss the distribution optimization problem in the supply chain based on time satisfaction rate maximized coverage the model and the corresponding heuristic algorithm are given.To verify the theoretical research results, the models and algorithms presented in this paper are implemented in the Weihai-HAIDU food industry information project as a typical application use case, in which product analysis, channel analysis and market analysis for the distribution network of supply chain. Application and the testing results are given and the theory proposed in the paper is verified.更多还原