漏磁信号处理中的经验模态分解和能量法阈值

The Empirical Mode Decomposition and Energy Threshold in Magnetic Flux Leakage Signal Processing

漏磁检测信号通常受多种噪声源污染。为了降低漏磁检测信号中的噪声,提高缺陷的识别效率,采用了经验模态分解方法,对漏磁信号进行去噪处理,来增强漏磁信号的信噪比。试验基于试样上的人工缺陷,漏磁信号被分解成若干固有模态函数(IMF)和一个残余分量,通过能量法选择最小能量的IMF作为阈值。采用该阈值作为IMF分量的选择依据,可以不依赖人为的经验判别。再将阈值后面的IMF分量相加,对漏磁信号进行重构。结果表明,采用该方法重构的漏磁信号的信噪比得到了较大的提高,同时滤波效果优于db3小波滤波。

The MFL signal is usually contaminated by various noises,in order to significantly reduce the noise signal in it,improving the efficiency of defect recognition,the empirical mode decomposition method was ultilized to de-noising for enhancing signal-to-noise ratio.Experiments were conducted on the pipeline steel samples with different depth artificial defects,The MFL signal was decomposed into several intrinsic mode functions（IMF） and a residual component,through the energy method selecting the IMF of minimum energy as the threshold which was the basis of the IMF components selected,and not relied on the experience of man-made judge.Then the magnetic flux leakage signal with the sum of IMF behind threshold was reconstructed.The result showed that the signal-to-noise ratio of the reconstructed MFL signal could be greatly enhanced using this method and filtering was better than db3 wavelet filter.