基于EEMD与FCM聚类的自动机故障诊断  

Automaton fault diagnosis based on EEMD and FCM clustering

作　　者：张玉学[1] 潘宏侠[1,2] 安邦[1] 

机构地区：[1]中北大学机械与动力工程学院,山西太原030051 [2]中北大学系统辨识与诊断技术研究所,山西太原030051

出　　处：《中国测试》2017年第3期106-110,共5页China Measurement & Test

基　　金：国家自然科学基金项目(51175480)

摘　　要：针对自动机使用中常见的故障检测与识别问题,考虑到自动机振动响应信号非线性、短时、瞬态和冲击特性,提出基于聚合经验模态分解(EEMD)和模糊C均值(FCM)聚类结合的自动机故障诊断方法。首先,使用EEMD分解方法对自动机的振动信号进行分解,结合相关系数提取前5个固有模态函数(IMF)分量的能量百分比作为特征值,再用模糊C均值聚类对特征值进行聚类分析。通过对自动机不同工况分别用EEMD和EMD方法进行故障分类识别对比,结果表明:所有样本的诊断结果与实际情况基本符合,证明EEMD法有更好的分类效果,分类正确率达93.75%。从而验证该方法能有效应用在自动机故障诊断中。A method automaton fault diagnosis based on ensemble empirical mode decomposition（EEMD） and fuzzy C means clustering （FCM） is proposed for the detection and identificationproblems of common faults when using automaton in view of the nonlinear, short-time, transientand impact properties of the automaton vibration response signal. Firstly, EEMD decompositionmethod is used to decompose automaton vibration signal, and energy percentage of components ofthe first five intrinsic mode function （IMF） is extracted as the fault feature value based on relevantcoefficients, then cluster analysis is carried out for those feature values based on FCM clustering.The EEMD and EMD methods are used to carry out fault classification, identification andcomparison according to different working conditions of automaton. The results show that thediagnosed results of all samples are basically conform to the actual conditions, with classificationaccuracy reaching 93.75%, which verifies that the method can be effectively used for automatonfault diagnosis.

关 键 词：自动机 聚合经验模态 模糊C均值聚类算法 能量比 故障诊断 

分 类 号：TP277[自动化与计算机技术—检测技术与自动化装置]