单通道AE信号盲分离的飞机构件监测方法研究  

Research on monitoring of aircraft structural components based on blindseparation of single channel AE signal

作　　者：何鹏举 刘刚易[2] 折廷廷 He Pengju;Liu Gangyi;She Tingting(Research ＆ Development Institute of Northwestern Polytechnical University in Shenzhen,Shenzhen 518057,China;School of Automation,Northwestern Polytechnical University,Xi＇an 710072,China)

机构地区：[1]西北工业大学深圳研究院,深圳518057 [2]西北工业大学自动化学院,西安710072

出　　处：《仪器仪表学报》2018年第8期34-42,共9页Chinese Journal of Scientific Instrument

基　　金：深圳市知识创新计划(JCYJ20170306154611415);西安市科技计划(2017086CG/RC049)项目资助

摘　　要：基于声发射（AE）技术的飞机结构件疲劳裂纹检测是飞机健康状态识别的一种有效方法。由于声发射信号的瞬态性、不确定性、微弱性和易受机电干扰性,使声发射检测技术很大程度上已演变成信号处理问题,目前多数研究报道的是单源声发射信号的降噪处理。然而,在实际应用中,不仅结构件出现疲劳裂纹时会产生AE信号,而且紧耦合的结构体之间也会因冲击载荷产生弹性波,以至观测信号一般是多源AE混合信号,波的传播时延的存在使得信号混合方式为卷积混合。针对目前测试方法不能正确识别AE信号,以致难以识别结构体是否存在疲劳裂纹的问题,提出一种具有信号源个数估计的单通道非负矩阵分解解卷积盲源分离算法。首先采用经验模态分解方法将单通道混合信号分解为多个本征模态函数;然后采用主成分分析法估计信号源个数,并重构观测信号;最后通过非负矩阵分解解卷积得到各个源信号。实验结果表明,单通道盲源分离算法能正确分离AE信号,为飞机关键结构件的疲劳裂纹监测提供了一种方法。It is an effective way to identify aircraft health status through the fatigue crack detection based on acoustic emission（ AE）technology. AE signal has the feature of transient,uncertainty,weakness and mechanical and electrical interference. To a large extent,AE detection technologyhas become the signal processing problem. At present,most related researchesfocus on the noise reduction of single source AE signal. However,in the practical application,not only structural parts＇ fatigue cracks,but also the elastic waves generated by the impact load among the tightly coupled structures produce AE signals. Therefore,the observed signals are usually multisource of AE mixtures. The existence of wave＇ s propagation delay makes the mixed mode becomeconvolutivemixture. AE signal cannot be correctly extracted by using current testing methods. It is difficult to identify whether there is fatigue crack. In this paper,a single channel nonnegative matrix factorization deconvolution blind source separation algorithm with source number estimation is proposed. First,empirical mode decomposition method is used to decompose the single channel mixed signal into multiple intrinsic mode functions. Then,the number of sourcesis estimated by principal component analysis. In this way,the observed signal is reconstructed.Finally,AE source signals are obtained through nonnegative matrix factorization deconvolution blind separation. Experimental results show that the single channel blind source separation algorithm proposed in this paper can correctly separate AE signals,which provide a method for fatigue crack monitoring of key structures of aircraft.

关 键 词：结构件疲劳裂纹 单通道AE检测 源个数估计 非负矩阵盲解卷积 

分 类 号：TP277[自动化与计算机技术—检测技术与自动化装置] TH17[自动化与计算机技术—控制科学与工程]