Nonlinear spectral correlation for fatigue crack detection under noisy environments

Vibration-based breathing crack identification using non-linear intermodulation components under noisy environment

“…Although both the researchers highlighted the sensitivity of spectral correlation for fatigue damage, the complete damage diagnostic methodology especially localization of breathing crack has not been suggested. However, the above work gives us further confidence and the motivation to propose a new reference-free breathing crack identification technique using spectral correlation [14][15][16][17][18][19][20] with vibration-based dynamic signatures.…”

“…Apart from this, the structure that exhibits predominantly stationary and linear dynamic response properties in its undamaged state tends to exhibit non-stationary and nonlinear properties once breathing crack sets in. [1][2][3][4][5][6][7] Therefore, the traditional spectral analysis which is better suited for stationary signals may not be very effective for the bilinear and non-stationary dynamic signatures [1][2][3]14,15 obtained typically from the structure with breathing crack.…”

“…spectral density function of the response) is generally used in most existing vibration-based breathing crack damage diagnostic techniques. However, Liu and colleagues 14,15 highlight that extracting all the sideband components is difficult using the traditional spectral analysis-based approaches, especially when the measured signals are corrupted with noise. It should be mentioned here that a structure or structural component with a breathing crack exhibits bilinear stiffness characteristics due to an instantaneous change in the state of the cracked domain from open to close, and vice versa.…”

“…Recent studies have shown that the methods like higher order spectral analysis of non-stationary and non-linear signals could provide results of interest for damage diagnosis. 14–20 Bouillaut and Sidahmed 16 have theoretically shown that the bilinear behaviour induced by the opening and closing behaviour of the breathing crack are basically cyclo-stationarity in nature and cyclic spectral analysis can be employed to extract damage-sensitive features from these cyclo-stationarity signals. A signal is defined as n -order cyclo-stationarity if its statistical moments until the n th order are periodically time-dependent; the frequency of such periodicity is called cyclic frequency.…”

“…[10][11][12][13] Furthermore, in the majority of cases, we can extract easily only the first few sidebands reliably using conventional spectral analysis, which becomes more difficult when the responses are heavily corrupted by measurement noise. 14,15 The damage diagnostic process can become much more effective if one can consider more sidebands exhibited by the structure due to breathing crack. This was clearly highlighted by Semperlotti et al 11 and Broda et al 6 This work attempts to consider more sidebands and helps in diagnosing smaller intensity breathing cracks present in the structure and warns when the damage is in its incipient stage, in order to take appropriate measures.…”