摘要
为了提高滚动轴承震动信号故障诊断的准确性,该文提出了一种基于小波包熵和聚类分析的集合型故障诊断方法。用小波包对滚动轴承振动信号进行三层分解,并提取其能量特征。以振动信号的能量分布作为概率分布进行信息熵运算,提取振动信号特征。为了检测是否有故障发生,结合减法聚类的思想,提出采用密度指标最高原则优化初始聚类中心的K均值聚类算法进行聚类。为了检验所提方法的有效性,采用不同故障直径的滚动轴承数据进行实验。实验结果表明,新的聚类方法克服了传统K均值聚类对初始聚类中心敏感的缺陷,其结果可以作为滚动轴承早期故障诊断的依据。
In order to improve the fault diagnosis accuracy of rolling bearing vibration signals,an ensemble approach based on wavelet packet entropy and clustering analysis is presented here.The method of wavelet packet is used to decompose rolling bearing vibration signals into three-layer,and extract the energy characteristics.The vibration signal energy distribution is used as the probability distribution to do the information entropy calculations and extract the vibration signal characteristics.To detect faults,combined with subtractive clustering,the K-means clustering method of optimizing initial cluster centers by the principle of highest density index is proposed.To test the effectiveness of the proposed method,the actual bearing data of rolling bearing with different fault diameters are provided in the experiment.The results show that the proposed approach avoids the sensibility of traditional K-means clustering to initial cluster centers and its result can be used as a basis for rolling bearing fault diagnosis.
出处
《南京理工大学学报》
EI
CAS
CSCD
北大核心
2013年第4期517-523,共7页
Journal of Nanjing University of Science and Technology
基金
国家自然科学基金(60974070)
辽宁省科学技术计划(2010222005)
关键词
小波包熵
减法聚类
滚动轴承
故障诊断
K均值聚类
wavelet packet entropy
subtractive clustering
rolling bearing
fault diagnosis
K-means clustering
