中文摘要：

为便于缺陷可视化和提高缺陷定量分析精度,必须降低超声检测过程中多种未知噪声源对超声A波信号的干扰。首先将超声A波信号进行相空间重构,得到一个相空间重构矩阵;然后进行基于FastICA算法的独立分量分析;最后从独立分量中提取出所需的超声信号。为了验证降噪效果,将该降噪方法与小波去噪方法进行对比。实验结果表明：该方法与小波去噪方法效果接近,极大地提高了超声信号的信噪比。同时,该方法与小波去噪方法相比具有自适应能力强、简单且易于实现等优点。

英文摘要：

In order to facilitate the flaw visualization and improve the flaw quantitative analysis precision, kinds of unknown noises should be eliminated to prevent interfering the ultrasonic A--wave during ultrasonic testing. Firstly,this method which obtained a RPS matrix used reconstrueted phase space to signals of A--wave ultrasonic. Secondly,ICA was applied to the RSP matrix based on FastI- CA algorithm. Finally, the available ultrasonic signals were obtained. To confirm the proposed method it was applied to de--noising ultrasonic signals of the test sample and was compared with the wavelet de--noising method. The experimental results show that the proposed method which enhances the signal- to--noise ratio enormously is very near to the wavelet de--noising method. The proposed method is easier to use and possesses a good adaptability.