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

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.