基于动态时间规整和自适应顺序形态变换分类的故障诊断方法

Fault diagnosis based on dynamic time warping and classification of adaptive rank-order morphological transform

研究了将动态时间规整(DTW)用于故障诊断的技术,使具有相同趋势但时序上并不完美一致的两序列之间距离最小化,发挥其在序列匹配中的重要作用。在前期研究提出的一种采用欧式距离度量两序列间的匹配程度的基于自适应顺序形态变换分类的故障诊断方法的基础上,由采用欧式距离变为采用DTW匹配来度量信号间误差,提出了三种用于故障诊断的新的自适应顺序形态变换分类方法。通过连续搅拌釜式加热器故障诊断问题的研究验证了所提出的三种新方法的有效性,并对比了它们与前期研究提出的方法的故障诊断效果。

The techniques of applying dynamic time warping （DTW） to fault diagnosis were studied to raise DTW＇ s im portant role in sequence matching by minimizing the distance between two sequences which display the same trends but do not perfectly align with each other. Based on the fault diagnosis method proposed in the earlier stage of the study that uses the classification of adaptive rankorder morphological transform and the Euclidean distance for measuring the matching degree between two sequences, the DTW matching was substituted for the Euclidean dis tance metric to measure the matching degree, and three new classification approaches for fault diagnosis were pres ented. The effectiveness of the three methods in diagnosis of faults was verified by the test of a continous stirred tank heater, and their performances were compared with that of method proposed in the earlier study.