Nonlinearity based signal processing for defect detection

When seeking out evidence for nonlinear behaviour, various signal processing techniques can be applied for the comparison of two signals, one being a slight distortion of the other. For instance, the pulse inversion technique compares the responses to two out-of-phase excitation signals. Alternatively, one can compare the response at a finite (nonlinear) excitation amplitude to a scaled response at a very low (linear) excitation, as performed in the scaling subtraction technique. In this report, several numerical as well as experimental examples are given in which these nonlinearity based signal processing techniques are used in practice to visualize damage features in solids. In view of kissing bond defect detection in friction stir welds, the pulse-inversion method was employed in a contact pitch-catch mode using a chirp signal. B-scan spectral heat maps obtained after pulse inversion allow to easily identify and size damage zones along the weld path. Secondly, the scale subtraction technique will be illustrated in combination with an ultrasonic sparse array SHM system to detect damage locations (impacts and delaminations in CFRP plates) without the need of baseline signals taken on an intact specimen. Finally, we show that the phenomenon of Local Defect Resonance (LDR) can be facilitated and validated using the scaling subtraction technique.status: accepte