摘要
针对样本熵和多尺度熵中相似性度量函数的突变问题,及它们在分析时间序列复杂性时捕捉不到高频组分信息的局限,提出了一种新的时间序列的复杂性度量方法——复合层次模糊熵(CHFE)。为了有效地提取滚动轴承早期故障特征,提出了一种基于CHFE、拉普拉斯分值和支持向量机的滚动轴承故障诊断方法。首先,提取振动信号的CHFE值;其次,采用拉普拉斯分值对特征向量进行降维优化;再次,建立基于支持向量机的多故障分类器,实现滚动轴承的故障诊断;最后,将该方法应用于实验数据分析,结果验证了方法的有效性。
Since the similarity measure in sample entropy and MSE changed abruptly and MSE might not capture high-frequency informations, a new method for measuring complexity of time series called CHFE was proposed. Meanwhile, in order to extract the early fault features of rolling bearings, a new fault diagnosis method was proposed based on CHFE, Laplace score for feature selection and support vector machine(SVM). First, the CHFEs were extracted from vibration signals of rolling bearing and then the Laplacian score was used to reduce dimension of features. Next, the SVM based multi-fault classifier was founded to fulfill the fault diagnosis. Finally, the proposed method was ap- plied to experimental data analysis and the results indicate the validity.
出处
《中国机械工程》
EI
CAS
CSCD
北大核心
2016年第15期2048-2055,共8页
China Mechanical Engineering
基金
国家自然科学基金资助项目(51505002)
安徽省高校自然科学研究资助重点项目(KJ2015A080)
关键词
多尺度熵
层次熵
复合层次模糊熵
滚动轴承
故障诊断
multi-scale entropy (MSE)
hierarchical entropy
composite hierarchical fuzzyentropy(CHFE)
rolling bearing
fault diagnosis