首页 | 官方网站   微博 | 高级检索  
     

转子共频相关故障源源数估计与子带盲分离
引用本文:李纪永,李舜酩,田国成,陈晓红.转子共频相关故障源源数估计与子带盲分离[J].振动.测试与诊断,2015,35(25):146-149.
作者姓名:李纪永  李舜酩  田国成  陈晓红
作者单位:(1.南京航空航天大学能源与动力学院 南京,210016)(2.山东中实易通集团有限公司 济南,250000)(3.南京航空航天大学理学院 南京,210016)
基金项目:航空自然科学基金资助项目(2012ZD52054);国家自然科学基金资助项目(61403193);南航基本科研业务费科研资助项目(NS2014081)
摘    要:针对转子异常振动产生含交叉频率的响应,其共频相关故障源不满足统计独立要求,提出利用非负矩阵法在频域中计算故障源个数,不计及源信号和混合系统特性,可以正确估计出故障源数目或源数上限。提出利用小波包分解故障信号,选择互信息较小的子带进行重构,剔除共频信号并进行盲分离,得到独立非相关的源信号,保留了故障信息。理论及实验结果证明了所提出方法的有效性。

关 键 词:相关源    非负矩阵分解    互信息    子带分解    盲源分离

Correlated Jointly Frequency Rotor Fault Sources Number Estimation and Sub-band Blind Separation
Li Jiyong,Li Shunming,Tian Guocheng,Chen Xiaohong.Correlated Jointly Frequency Rotor Fault Sources Number Estimation and Sub-band Blind Separation[J].Journal of Vibration,Measurement & Diagnosis,2015,35(25):146-149.
Authors:Li Jiyong  Li Shunming  Tian Guocheng  Chen Xiaohong
Abstract:Crosscover frequency response is produced by the rotor under abnormal vibration, so rotor jointly frequency fault sources cannot be satisfied statically independent request. The traditional source number estimation such as singular value decomposition method and standard independent standard analysis cannot extract fault sources. This paper estimates fault sources number with a non-negative method in the frequency domain, source and mixture system characteristics not in consider, and then to extract fault signals with wavelet packet, and separate recovered signals with small mutual information, to eliminate jointly frequency signal, to obtain independent non-correlated sources. The feasibility is verified by practical and theoretical methods.
Keywords:correlated sources  non-negative matrix factorization  mutual information  sub-band decomposition  blind source separation
点击此处可从《振动.测试与诊断》浏览原始摘要信息
点击此处可从《振动.测试与诊断》下载全文
设为首页 | 免责声明 | 关于勤云 | 加入收藏

Copyright©北京勤云科技发展有限公司    京ICP备09084417号-23

京公网安备 11010802026262号