基于模糊聚类算法的故障数据分析与类型识别

Fault Data Analysis and Type Recognition Based on Fuzzy Clustering Algorithm

为精确分析测量系统故障数据和识别故障类型,提出一种基于模糊聚类算法的故障数据分析方法。该方法首先用小波变换有效地检测出系统故障的微弱非线性不规则信号,再用模糊聚类的方法对故障进行分类识别。由于该算法在目标函数中加入隶属度函数,同时定义明可夫斯基的距离测度,因此能够克服K-means算法不适用于进行非凸形状的聚类的缺点,从而使诊断的数据更加精确。

To accurately analyse the fault data of the measurement system and identify the type of the fault, presents a fault diagnosis approach based on fuzzy clustering algorithm. Firstly, uses the wavelet transform to locate the weak and nonlinear signal when there are faults, and then uses the fuzzy clustering algorithm to identify the type of the fault. Because the membership degree and the Minkowski distance is defined in the objective function, the proposed scheme can overcome the disadvantages of clustering the non-convex data, so as to achieve the accurate fault diagnosis.