Ensemble empirical mode decomposition of impact-echo data for testing concrete structures

Ensemble empirical mode decomposition (EEMD) is investigated to decompose the impact echo (IE) testing data into different spectral composition for defect signal extraction. The effects of critical EEMD parameters are studied. As IE signals have strong surface waves with wide bandwidth, the amplitude of the added white noise is larger than that normally used for other applications to successfully decompose all interested modes. The number of ensemble trials increases with increasing noise amplitude. The influence of the frequency of the signal to be extracted on the EEMD performance is also analyzed. The results show that the high frequency resonance mode is easier to be extracted than the low frequency resonance mode from the IE signal. The effectiveness of the EEMD method for IE signal decomposition is demonstrated using both numerical simulations and experimental tests.

► EEMD is investigated for decomposing impact echo (IE) signals.
► Large noise amplitude is required for EEMD of IE signals with strong surface waves.
► The effect of IE mode signal frequency on EEMD performance is analyzed.
► EEMD performs better than EMD for decomposing IE signals with strong surface waves.