模糊WINGS视角下的ANP加权矩阵新构造方法  

New construction method for ANP cluster-weighted matrix from view of fuzzy WINGS

作　　者：孙永河[1] 李春好[2] 谢晖[1] 段万春[1] 

机构地区：[1]昆明理工大学管理与经济学院,昆明650093 [2]吉林大学管理学院,长春130025

出　　处：《计算机工程与应用》2014年第12期26-32,37,共8页Computer Engineering and Applications

基　　金：国家自然科学基金(No.71261013;No.71371083);教育部人文社会科学研究青年基金项目(No.10YJC630218);云南省科技计划项目(No.2010ZC060);云南现代化管理与新型工业化研究基地前期研究课题(No.YNXXJD201102)

摘　　要：构造因素集加权矩阵是ANP系统未加权超矩阵到加权超矩阵转化的一个关键技术。然而从已有的三种构造方法看,两两比较法比较机理混乱,而等权矩阵假设构造法通常是无效的,并且基于DEMATEL(决策试行与评价实验室)的构造方法不仅存在忽视因素集自身强度的内在不足,而且也难以反映专家在对因素集之间影响关系判断时存在的"不精确性"。为克服上述缺陷,提出一种模糊WINGS(加权影响情景下的非线性测度体系)视角下的ANP因素集加权矩阵新构造方法。该方法一方面给出改进后的模糊DELPHI决策程序,充分考虑了专家判断过程中的"不精确性"。另一方面,系统提出模糊WINGS的方法思路,在系统因素集影响关系判别时充分考虑了因素集的自我影响强度,使因素集直接影响矩阵的构造更为合理。实例对比验证结果表明,该方法是科学合理的,有着较强的实践应用可操作性。It is vital to construct a Cluster-Weighted Matrix（CWM）among clusters when converting an unweighted super-matrix of ANP（Analytic Network Process）system to the weighted supermatrix. However, existing three construction methods have the following disadvantages. First, pairwise comparison method has disordered comparison mechanism. Second, the method of supposing equal weight matrix is usually inefficient unless the CWM is a nominal matrix. Third, the construction method based on DEMATEL（Decision Making Trial and Evaluation Lab）not only ignores the Impor-tance of Each Cluster（IEC）, also is the inaccuracy of expert judgments overlooked. To overcome the above drawbacks, this paper suggests a new fuzzy WINGS（Weighted Influence Non-linear Gauge System）to construct CWM of ANP. On one hand, in the new approach, the inaccuracy of expert judgments is considered by improved fuzzy DELPHI procedures. On the other hand, the idea of fuzzy WINGS is suggested in detail, and IEC is added when analyzing the influence rela-tion among clusters, which makes the direct influence matrix be more rational. By a numerical case, the approach is vali-dated to be scientific and it can be well applied to solving the real world issues.

关 键 词：模糊加权影响情景下的非线性测度体系(WINGS) 网络分析法 超矩阵 加权矩阵 

分 类 号：N94[自然科学总论—系统科学]