Experimental Ultrasonic NDT Signal of Steel Based on Improved Empirical Mode Decompositions

Defects localization and materials characterization are the main goals in ultrasonic NDT. Because of the nonuniform propagation and noisy environment, received echoes are nonstationary, nonlinear and formed of multiple overlapped components. To improve the localization of echoes, several algorithms are proposed in the literature. In this paper, the algorithm based on the complete ensemble empirical mode decomposition with adaptive noise (CEEMDAN) has been evaluated and compared for ultrasonic applications. Steel plate with a defect is used to evaluate the effectiveness of the method for ultrasonic signals. The experiment has been performed using the low-frequency ultrasonic system conducted at the NDT Laboratory of Jijel University. Numerical simulation and experimental tests show that this method is complete, with a numerically negligible error. The results show that, compared with variants EMD, CEEMDAN also provides a better spectral separation of the modes and a lesser number of sifting iterations is needed, reducing the computational cost. Experimental results clearly exhibit that the combined CEEMDAN and Hilbert Huang Transform (HHT) is an effective processing tool to analyze ultrasonic signals for defect detection and characterization.