新产品开发项目组织聚类与流程优化研究

发布时间：2018-01-24 01:29

本文关键词： 新产品开发项目管理设计结构矩阵(DSM) PageRank算法信息熵谱聚类　出处：《北京科技大学》2017年博士论文　论文类型：学位论文

【摘要】：在当前经济全球化不断加剧、市场竞争日益激烈的新形势下,为了赢得市场,企业必须以客户需求为导向,不断快速地推出高质量的新产品。新产品开发项目是由“产品-流程-组织”三个领域高度耦合构成的复杂系统架构,产品的需求分析是影响产品开发成功与否的首要因素；组织优化是降低管理复杂性,保证产品开发顺利进行的有效途径；流程优化是缩短项目开发时间,降低开发成本的主要手段。迄今为止,关于这些方面的研究已经取得了许多成果,但在理论方法和实际应用方面尚存在一些不足。本文基于设计结构矩阵理论(DSM)等方法,以企业中产品开发的的实际情况为背景,深入分析新产品开发项目中流程和组织的特点,主要研究内容如下：(1)准确预测用户对产品的功能需要,以及构建结构合理的项目开发组织是企业生存的关键。因此,本文开展了基于客户需求的新产品开发组织结构优化研究。首先,提出建立网络平台下的客户体验网络,基于用户的网络浏览数据建立产品功能网络,在此基础上,利用PageRank算法挖掘客户对新产品的功能需求。然后,为了优化组织结构,从信息熵所描述的组织有序度的角度,提出了组织DSM信息熵概率的计算方法,并建立了组织DSM的类内、类外和整体的信息熵计算模型。进一步,提出了基于信息熵的组织DSM二阶段聚类算法来优化组织结构,降低组织复杂性。最后,通过案例分析,验证了本文提出模型的有效性。(2)为了弥补传统DSM聚类算法稳定性差的缺点,本文开展了基于谱聚类的新产品开发项目组织优化研究。首先,结合新产品开发组织中的技术信息传递的特点和团队在目标实现中的不同作用,提出采用PageRank算法来识别组织核心团队。其次,根据杰卡德相似性系数的思想,结合团队之间的技术信息沟通情况,建立了计算团队之间相似度的数学模型,并构建相似性矩阵。进一步,给出了组织DSM谱聚类过程,并将在PageRank算法中得分靠前的团队作为谱聚类的初始聚类中心。最后,通过案例分析,与现有的两种DSM聚类算法做对比,结果显示本文提出的谱聚类方法显示出了更好的聚类效果和更强的鲁棒性。(3)为分析产品开发过程中的迭代对研发项目时间和成本的影响,本文开展了基于返工风险传播和预处理的新产品开发项目流程DSM优化研究。首先,研究了返工变更的传播过程,在直接返工概率和间接返工概率的基础上建立了返工因子设计结构矩阵,量化了返工变更传播和返工时间、范围对返工风险的影响。然后,提出采用“返工预处理”的方法降低返工影响强度,构建了预处理后的返工影响强度计算模型。进一步,基于返工预处理和返工因子DSM构建了项目净收益模型,并以项目净收益最大化作为流程DSM优化目标。最后,通过案例验证了本文提出模型的有效性,优化与仿真结果表明采用本文提出的方法可有效控制返工风险。
[Abstract]:In the current economic globalization intensifies, the new situation of increasingly fierce competition in the market, in order to win the market, the enterprise must take the customer demand as the guidance, the rapid introduction of new products of high quality. The new product development project is a complex system architecture consists of three areas of "product flow organization" highly coupled, demand analysis the product is a primary factor affecting the success of product development; organization optimization is to reduce the complexity of management, effective way to ensure the smooth development of the products; process optimization is to shorten the time of project development, the main means of reducing development costs. So far, the Research on these has made many achievements, but in theory, method and practice the application still exist some shortcomings. This paper based on the theory of design structure matrix (DSM) method, based on the actual situation of product development of enterprises in the background of deep. The analysis of the characteristics of process and organization of new product development project. The main contents are as follows: (1) the accurate prediction of the user of the product function, and the key to construct reasonable project development organization is the survival of the enterprise. Therefore, this paper carried out the research on the structure optimization of new product development organization based on customer demand. First, put forward to establish a network platform customer experience network, users browsing data to establish the product function based on network, on the basis of this, using the PageRank algorithm for mining customer functional requirements for new products. Then, in order to optimize the organizational structure, order degree entropy described from the perspective of the organization, puts forward the calculation method of DSM entropy probability the organization and the establishment of the DSM class, class of calculation model and overall information entropy. Further, put forward to optimize the two stage DSM clustering algorithm based on information entropy Organization structure, reduce the complexity of organization. Finally, through case analysis, to verify the validity of the proposed model. (2) in order to make up for the traditional DSM clustering algorithm, the shortcoming of poor stability, this paper carried out the research on the new product development project organization optimization based on spectral clustering. First, different combination of features and new product information technology team the development of tissue transfer in the realization of the goal of the proposed PageRank algorithm is used to identify the organization's core team. Secondly, according to the Jaccard similarity coefficient method, combined with the status of information communication technology situation between the team, establishes a mathematical model for calculating the similarity between the team, and construct similarity matrix. Further, given DSM spectral clustering process, and the scores on the team in the PageRank algorithm as the initial cluster center of spectral clustering. Finally, through the case analysis, and the existing two kinds of poly DSM Comparison of algorithms, results show that the spectral clustering method is proposed in this paper shows better robustness and stronger clustering effect. (3) in order to analyze the effect of iteration in the process of product development for the project time and cost, this paper carried out the research of new product development project process optimization DSM rework risk propagation and Pretreatment Based on the study of the propagation process. Firstly, rework changes, based on direct and indirect rework rework probability probability is established on the rework factor design structure matrix, quantitative change propagation and rework, rework time, influence range of rework risk. Then, the method of "rework pretreatment" reduce rework impact strength. The calculation model is constructed after pretreatment of the rework impact strength. Further, rework pretreatment and rework factor DSM to construct the project model based on net income, and net income to project into Finally, a case is given to verify the effectiveness of the proposed model. The optimization and simulation results show that the risk of rework can be effectively controlled by the proposed method in this paper, DSM.

【学位授予单位】：北京科技大学
【学位级别】：博士
【学位授予年份】：2017
【分类号】：F273.2

【相似文献】