AHP中判断矩阵一致性修正的模式识别法

基于模式识别原理，给出了判断矩阵一致性修正的一种方法。在该方法中，首先把由专家或决策者给出的判断矩阵Ａ转化为ｎ个具有完全一致性的正互反阵。再将这ｎ个正互反阵和Ａ分别转化为反对称阵Ｂ（ｉ），ｉ＝１，…，ｎ和Ｂ。在Ｂ（ｉ），ｉ＝１，…，ｎ生成的线性空间中寻找与Ｂ最贴近的正互反阵Ｂ～＊，将Ｂ～＊对应的满足完全一致性的正互反阵Ａ～＊作为Ａ的修正矩阵，同时得到了Ａ～＊对应于最大特征值ｎ的正特征向量的计算公式。

The Pattern Recognition Method for Correcting the Judgement Matrix into One with Complete Uniformity in AHP

Basing on the Principle of pattern recognition, we present a new method for correcting the judgement matrix into one with complete uniformity.In the method,we obtain n pieces of reciprocal matrices and transform them into inverse symmetric matrices respectively.An inverse symmetric matrix in the linear space expanded by those n pieces of inverse symmetric matrices is obtained and be used as the correction one of the inverse symmetric matrix corresponding to the judgement matrix.