Denoising Method for Ultrasonic Flaw Echo Based on EMD and Local Entropy

Aiming at the challenge of eliminating structural noise and electronic noise in ultrasonic nondestructive testing when detect metallic materials, a denoising method combining empirical mode decomposition (EMD) with local entropy is proposed for flaw echo denosing. In this method, The Intrinsic Mode Functions (IMFs) components and residual components of ultrasonic signal were obtained through the EMD method. A Similarity Measure method were used to select the signal-dominant modes. Then, the local entropy of the first signal-dominant mode was calculated to identify the location of defect echoes, and a window was used to preserved defect echoes in the all relevant modes. Finally, the signals after denosing was obtained by accumulating defect echoes and residual components. Denoisng results of simulative signal and experimental data show that the presented method can significantly remove noise from ultrasonic detecting signal and keep the flaw echo to the maximum extent.