基于BP神经网络的AHP判断矩阵调整方法  

A method of comparison matrix adjustment based on BP neural network

作　　者：于建华[1] 孙首群[1] 杨凡[1] 满微微[1] 

机构地区：[1]上海理工大学机械学院,上海200093

出　　处：《现代制造工程》2011年第3期13-17,共5页Modern Manufacturing Engineering

基　　金：国家自然科学基金资助项目(50875174);国家863计划资助项目(2009AA11Z211);上海市教育委员会重点学科建设项目(J50503)

摘　　要：提出了一种基于BP神经网络的判断矩阵一致性调整方法。该方法在建立判断矩阵一致性调整BP神经网络模型的基础上,通过对原始判断矩阵施之一定的数学变换,构造出完全一致性判断矩阵并将其作为BP神经网络的教师信号。经BP神经网络算法的调整使原始判断矩阵逐渐逼近构造的完全一致性判断矩阵,从而达到一致性指标要求。计算实例表明,此种方法是可行及可靠的。A new method based on BP neural network to adjust the consistency of comparison matrix was presented. A BP network model was established. A completely consistent comparison matrix is constructed by mathematical transform of the original comparison matrix which was regarded as the teacher signal of the BP neural network. The original comparison matrix is converted to an approximation of the completely consistent matrix by the BP network algorithm adjustment, so as to satisfy the consistency index requirements. Numerical examples indicate that this method was effective and reliable in practice.

关 键 词：BP神经网络 层次分析法 判断矩阵 完全一致性矩阵 

分 类 号：TP399[自动化与计算机技术—计算机应用技术]