Performance of Vibration Based Damage Detection Algorithms for Detection of Disbond in Stiffened Metallic Plates

Though vibration based health monitoring has been the focus of attention for quite some time, there exist a strong requirement for an extensive and comprehensive study on the relative performance of different damage detection algorithms under different damage scenarios. To fulfill this objective, a stiffened aluminium plate has been selected to make a comparative study on the performance of several vibration based damage detection techniques, namely, Modal Curvature, Gapped Smoothing Method/Modal Curvature, Generalized Fractal Dimension and Uniform Load Surface Curvature based on four different criteria: variation of damage intensity (i.e., disbond length), position of stiffener and disbond, the effect of noise and capability to detect multiple damage. In addition to this, a new approach, based on the curvature of wavelet coefficients, has been presented. It is found that this novel approach is extremely effective in determining the presence and location of damage under different situations. The entire numerical modelling is done in ANSYS 14.0 and the damage detection algorithms written in MATLAB codes have been used to generate the required damage indices using the modal data retrieved from ANSYS. The study ultimately enlightens inherent characteristics of the various damage detection algorithms under different damage conditions.