Ultrasonic signal identification by empirical mode decomposition and Hilbert transform

A novel approach is proposed to identify the arrival time of overlapping ultrasonic echoes in time-of-flight diffraction (TOFD) flaw detection. The nonlinear and nonstationary ultrasonic signal is decomposed using empirical mode decomposition method to obtain intrinsic mode functions, which are used for signal reconstruction. By applying Hilbert transform to the reconstructed signal, the arrival time of each echo can be clearly identified. The efficiency and feasibility of this approach in enhancing the time resolution of ultrasonic signal are validated by simulation and experiment, where 98% of flaws in 12-, 10-, and 8-mm-thickness pipelines can be identified with an average accuracy of 0.2 mm. The blind area of TOFD has been reduced to 2.5 mm under the surface. The frequency and angle of the probe pair have little influence on the identification.