基于模糊优先关系矩阵的系统评价方法

为处理系统评价中各评价指标的一致无量纲化问题,避开模糊综合评价方法中建立隶属度函数的困难,探讨了用模糊优先关系矩阵A的优度值作为各评价指标的一致化和无量纲化值的新途径。为充分利用A的一致性信息和提高A的优度值计算结果的可信程度,提出了A的最优模糊一致性判断矩阵、一致性指标函数和一致性指标临界值。研制了用加速遗传算法检验、修正A的一致性,并同时计算A各评价对象优度值的新的系统评价方法(AGA-FPRM)。理论和实例分析的初步结果表明,AGA-FPRM方法直观、实用,矩阵修正幅度较小,计算结果稳定、精度高,可在模糊层次分析法理论与实践中推广应用。

System Evaluation Method Based on Fuzzy Preferential Relation Matrix

In order to make consistency and to eliminate dimensions of system evaluation indexes, and to avoid determining membership function in fuzzy comprhensive evaluation, a new approach is studied with which preference values of fuzzy preferential relation matrix A can be used as consistency and non-dimension evaluation indexes.Optimal fuzzy consistency judgement matrix, consistency index function and consistency index critical value of matrix A are presented in order to make the most of consistency information of matrix A ,and to improve believable degree of computed preference values of matrix A.A new system evaluation method, named AGA-FPRM, is proposed to check and correct the consistency of matrix A and to compute preference values at the same time by using accelerating genetic algorithm improved by the authors. The results of theoretical analysis and case study show that AGA-FPRM is visual, practical, that correcting range of matrix A less, that its result is both stable and highly precise, and that it possesses important theoretical significance and broad application value in fuzzy analytic hierarchy process.