%0 Journal Article
%T 基于融合距离法的柴油机故障诊断方法
%A 刘
%A 罡
%A 王兴成
%A 杨昺崧
%J 大连海事大学学报
%D 2017
%X 针对故障特征变量在相关性模糊时,经典马氏距离和欧氏距离均无法有效计算故障特征的问题，将马氏距离与欧氏距离相融合，提出一种新的距离融合度量方法.该方法利用特征变量相关系数来确定权系数，将马氏距离与欧氏距离进行动态加权，兼顾了特征变量的相关性和独立性，可有效地提高故障诊断精度.仿真算例分别从诊断精度和聚类效果上验证了融合距离法的有效性.</br>In the case of unclear correlation of faults characteristic variable by a complex of monitoring information of diesel engine, the effective features of faults cannot be calculated by neither through Mahalanobis distance nor though Euclidean distance. However, a distance measurement method named fusion distance was put forward. The weight coefficients were determined by using characteristic variable correlation coefficient, and the dynamic weighting of Euclidean distance and Mahalanobis distance was carried out under consideration of both the correlation and independence of characteristic variables, which can effectively improve the accuracy of fault diagnosis. The validity of the fusion distance method was verified by the diagnostic accuracy and clustering effects respectively.
%K 内蒙古自治区高等学校科学研究项目（NJZZ16178）．
%U http://xb.dlmu.edu.cn/CN/abstract/abstract414.shtml