Research on Ultrasonic Damage Detection and Noise Reduction of Concrete Structures Based on CEEMDAN and SIGMOD Smoothing Threshold Methods

Résumé fourni par la source

In this study, a simulation environment for concrete ultrasonic inspection was constructed using COMSOL Multiphysics software, and a 16-probe ultrasonic phased array was used to acquire signals from defects, and the acquired defect signals were subjected to complete set empirical mode decomposition (CEEEMDAN) processing. The signal is decomposed into multiple intrinsic modal functions (IMFs) by CEEEMDAN, and the effective IMFs, noise-containing IMFs components are filtered according to the energy contribution. A new sigmoid function is introduced as the wavelet threshold function, and its smoothness is utilized to make the signal coefficients transition smoothly, and for the problem of its large sensitivity near the threshold, the exponential part of it is subjected to 1/2 dimensionality reduction, which increases the robustness of signal processing. Finally, the decomposed noise-containing IMFs components are reconstructed with the wavelet transform after improving the threshold function. The results show that the signal after joint noise reduction by CEEMDAN and Sigmoid smoothing threshold function is effectively improved in terms of signal-to-noise ratio (SNR) and mean square error (MSE) relative to the traditional noise reduction methods.