结合小波变换和稀疏主成分分析的复合材料缺陷信号增强算法  

Enhancement of Defect Signals in Composite Materials Using a Combination of Wavelet Transforming and Sparse Principal Component Analysis

作　　者：张圆 刘薇 李津蓉[1] 孙勇智[1] ZHANG Yuan;LIU Wei;LI Jinrong;SUN Yongzhi(School of Automation and Electrical Engineering,Zhejiang University of Science&Technology,Hangzhou Zhejiang 310023,China)

机构地区：[1]浙江科技学院自动化与电气工程学院,浙江杭州310023

出　　处：《传感技术学报》2022年第12期1664-1670,共7页Chinese Journal of Sensors and Actuators

基　　金：国家自然科学基金面上项目(62173306);浙江省教育厅科研项目(Y202044842)。

摘　　要：脉冲热成像技术被广泛应用于碳纤维复合材料内部缺陷检测。由于原始热成像数据包含不均匀背景及检测噪声,缺陷信号的可视化效果较差,无法直接进行缺陷检测及识别。针对上述问题,提出结合小波变换和稀疏主成分分析(Wavelet Transforming and Sparse Principal Component Analysis,WT-SPCA)的特征提取方法,以提高缺陷信号的可视化效果。首先利用小波变换进行噪声信号去除,进一步采用稀疏主成分分析提取缺陷信号特征。实验结果表明,WT-SPCA方法可有效去除不均匀背景及噪声干扰,准确提取缺陷特征。与主成分分析、稀疏主成分分析等特征提取方法相比,WT-SPCA能够有效提高缺陷信号的可视化效果及缺陷区域的信噪比水平。Pulsed thermography is widely used for detecting the subsurface defects inside carbon fiber reinforced polymers.However,it is difficult to identify defects direct from the original thermal images,since the defect signals are heavily masked by the non-uniform background and measurement noise.In order to enhance the visualization of defects,a feature extraction method based on wavelet transforming and sparse principal component analysis(WT-SPCA)is proposed.Firstly,wavelet transforming is applied to reduce measurement noise from original thermal images.Then,SPCA is adopted to extract features of the defects from the filtered thermogram.Finally,the defect signals can be visualized through the loading distributions of the features.The experimental result illustrates that WT-SPCA can effectively remove the interference of uneven background and measurement noise,such that the features of the defects can be extracted more accurately.In comparison with the methods of principal component analysis and sparse principal component analysis,WT-SPCA significantly improves the visualization of defects as well as the level of signal-to-noise ratios for the defective areas.

关 键 词：脉冲热成像 特征提取 小波变换 稀疏主成分分析 碳纤维复合材料 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]