基于压缩感知技术的旋转机械碰摩声发射信号压缩  

Rubbing Acoustic Emission Signal Compression of Rotating Machinery Based on Compressive Sensing

作　　者：秦康[1] 邓艾东[1] 张红星[2] 唐标[2] 颜喜[1] 

机构地区：[1]东南大学火电机组振动国家工程研究中心,江苏省南京市210096 [2]江苏苏美达集团公司,江苏省南京市210018

出　　处：《中国电机工程学报》2013年第S1期160-165,共6页Proceedings of the CSEE

基　　金：国家自然科学基金资助项目(51075068)~~

摘　　要：针对进行旋转机械碰摩故障诊断时由于声发射信号数据量过大而导致节点数据传输速率低下的问题,引入压缩感知技术对碰摩声发射信号进行压缩。根据压缩感知理论,该文运用高斯随机矩阵作为测量矩阵对碰摩声发射信号进行压缩,并用正交匹配追踪算法作为重构算法对压缩结果进行重构,同时与原信号相比较得到重构误差。选取测量矩阵(M?N维)中不同的M值,分别对信号进行压缩和重构,通过分别比较压缩和重构的结果,研究M值的选取对于运用该技术时压缩和重构效果的影响。实验结果表明,该技术可以有效地对碰摩声发射信号进行压缩,并能以较小的误差对信号进行重构。同时,在选取测量矩阵的M值时,必须对压缩的效率和重构的精度进行综合考虑,以达到最好的效果。To avoid the problem of low transmission rate brought about by the large acoustic emission data volume in rotating machinery when diagnosing rubbing fault, the compression of rubbing acoustic emission signal based on compressive sensing was proposed. According to compressive sensing, Gaussian matrix was used as measurement matrix to compress the signal, orthogonal matching pursuit algorithm was used as reconstruction algorithm to reconstruct the compression result, and then the reconstruction result was compared with the original signal to get the reconstruction error. The signal was compressed and reconstructed respectively when choosing different M of the measurement matrix(M?N).The influence of the choice of M on the result of the compression and the reconstruction was researched by comparing the result of the compression and the reconstruction respectively. The experiments indicate that by using this technology the rubbing acoustic emission signal can be effectively compressed, and the compression result can be reconstructed with small error. Both the efficiency of compression and the accuracy of reconstruction should be taken into consideration when choosing the M of the measurement matrix to get the best result.

关 键 词：声发射 碰摩 信号压缩 压缩感知 正交匹配追踪算法 

分 类 号：TH165.3[机械工程—机械制造及自动化]