脉冲远场涡流检测PCA-ICA联合消噪技术  

PCA-ICA integrated technique for noise suppression of pulsed remote field eddy current inspection

作　　者：刘相彪[1] 李勇[1] 陈振茂[1] 闫贝[1] 李一力[1] 

机构地区：[1]西安交通大学航天航空学院机械结构强度与振动国家重点实验室核能结构安全检测与完整性评价研究中心,陕西西安710049

出　　处：《传感器与微系统》2015年第1期69-72,共4页Transducer and Microsystem Technologies

基　　金：国家自然科学基金资助项目(51007069/E070104)

摘　　要：脉冲远场涡流作为一种新兴的电磁无损检测技术,国内外研究发现其对铁磁性管道检测具有显著的优势。然而在实际检测中,由于磁导率不均等管道自身材质因素的影响,使检测信号信噪比较低,易造成缺陷误判。针对这一问题,提出了主成分分析—独立分量分析(PCA-ICA)联合消噪技术。单通道缺陷扫查信号经过PCA处理后,利用前二阶主成分构建虚拟双通道信号作为ICA输入矩阵,采用扩展特征矩阵联合近似对角化(FJADE)算法实现在单通道信号欠定条件下缺陷信号与噪声信号的分离,达到降噪目的。通过仿真与实验研究证明:PCA-ICA技术可以有效降低磁导率不均等噪声信号对检测结果的影响,提升了检测结果的信噪比。As one of the advanced electromagnetic nondestructive evaluation technique,pulsed remote field eddy current （PRFEC） has been found advantageous in inspection of magnetic tubular structures （ MTS ）. However, inhomogeneous distribution of permeability of practical MTS results in noise in PRFEC signals, and thus decreases the signal-to-noise ratio （ SNR ） and evaluation accuracy. In light of this, propose principal component analysis- independent component analysis （PCA-ICA） integrated technique for suppression of noise in PRFEC signals. After signals are processed via PCA, the 1-order and 2-order vectors are input into ICA, flexible joint approximative diagnalization of eigenmatrix （ FJADE ） algorithm is subsequently applied to distinguish the defect signals and noise. Through simulations and experiments, it has been found the proposed PCA-ICA integrated technique is capable of effectively suppressing the noise caused by iuhomogeneous distribution of permeability of MTS, and thus enhances SNR of PRFEC.

关 键 词：脉冲远场涡流检测 主成分分析 独立分量分析 铁磁性管道 磁导率不均 

分 类 号：TG115.28[金属学及工艺—物理冶金]