基于遗传算法的约束等式修正法  

A modified method of reducing equality based on genetic algorithm

作　　者：王佳豪 李岩 王利民[1] 邓全才[1] 郭雅静 WANG Jiahao;LI Yan;WANG Limin;DENG Quancai;GUO Yajing(Hebei University of Architecture,Zhangjiakou,Hebei 075000)

机构地区：[1]河北建筑工程学院,河北张家口075000

出　　处：《河北建筑工程学院学报》2024年第1期258-262,共5页Journal of Hebei Institute of Architecture and Civil Engineering

摘　　要：层次分析法研究中的热点问题之一是判断矩阵的一致性修正问题。约束等式修正法可以快速将判断矩阵修正为一致性矩阵,该方法对矩阵的数值改动较大,对专家意见的保留度极低,使矩阵丢失原本含义,降低层次分析法的可靠性。在约束等式修正法中引入遗传算法的思想,通过构建合理的适应度函数,既能使矩阵通过一致性检验,又可以最大限度地保留专家意见。通过对适应度函数参数值的调整,可以自由调节专家意见与矩阵一致性指标的权重从而解决各种场景下的判断矩阵修正问题。本方法适应范围广且修正稳定。在解决某企业零件寿命预测的问题中发挥了重要的作用,有一定的实用价值。One of the hot issues in analytic hierarchy process(AHP)research is the consistency correction of judgment matrices.The constraint equation correction method can quickly revise the judgment matrix to a consistency matrix.However,this method often leads to significant numerical changes in the matrix,resulting in a very low retention of expert opinions and causing the matrix to lose its original meaning,thereby reducing the reliability of the AHP.This paper introduces the concept of genetic algorithms into the constraint equation correction method,and by constructing a reasonable fitness function,it can ensure that the matrix passes the consistency test while maximizing the retention of expert opinions.By adjusting the parameter values of the fitness function,one can freely adjust the weighting between expert opinions and matrix consistency indicators,thus solving the judgment matrix correction problems in various scenarios.This method has a wide applicability and stable correction performance.It has played an important role in solving the problem of predicting the lifespan of parts in a certain enterprise and has practical value.

关 键 词：层次分析法 约等式修正法 遗传算法 

分 类 号：TP311[自动化与计算机技术—计算机软件与理论]