基于自适应低秩稀疏分解的超声缺陷回波检测方法

Ultrasonic Defect Echo Identification Based on Adaptive Low-Rank Sparse Decomposition

超声无损探伤在金属材料微小缺陷检测中有着广泛的应用,但采集的回波信号通常受到噪声干扰甚至完全被噪声掩盖,为了辨别被噪声干扰的缺陷反射信号,提出了一种基于自适应低秩矩阵分解的超声缺陷回波检测方法。首先对原信号进行短时傅里叶变换并提取幅度谱和相位谱,引入基于误差重建的背景矩阵秩估计方法,用于估计低秩稀疏分解所需的低秩度参数。然后通过低秩稀疏分解将幅度谱分解为低秩、稀疏和噪声三部分,舍弃噪声部分。最后使用时频掩蔽分离出缺陷信号幅度谱并运用逆短时傅里叶变换获得回波信号。用本文提出的方法分别对仿真和实测信号进行处理,结果表明该方法在缺陷回波检测方面是有效的。

Ultrasonic nondestructive detection has been widely used in minor defect of metallic materials detecting,but the echo signal is always disturbed by noise.In order to identify the defect echo contaminated by noise,a method to detect defects based on adaptive low-rank sparse decomposition is proposed.Firstly,the short time Fourier transform(STFT)is performed on the original signal to obtain the amplitude and phase spectra,and a rank approximation method based on error reconstruction is adopted to estimate the parameter required by the decomposition algorithm.Then,through the low-rank sparse decomposition,the amplitude spectra is decomposed into the low-rank,sparse and noise part,and the noise part is discarded.Finally,the binary time-frequency mask is used to fuse the low-rank and sparse part,in which the amplitude spectra of defect signal is identified.Through the inverse STFT,the amplitude spectra could be transformed to the time-domain,and the corresponding defect echo is acquired.Experimental results show the effectiveness of proposed method in defect echo identification.