Performance evaluation of damage detection algorithms for identification of debond in stiffened metallic plates using a scanning laser vibrometer

Abstract

While several studies have focused on the detection and localization of delamination in composite plates, few comprehensive studies have been performed for the identification of debond in stiffened metallic plates using vibration-based approaches. Therefore, this study is motivated by the need to evaluate the qualitative performance of existing damage detection algorithms, namely modal curvature, the gapped smoothing method (GSM), the generalized fractal dimension (GFD) and the wavelet transform coefficient (WTC), in detecting debond in stiffened metallic plates. Extensive experimental investigation is performed using laser Doppler vibrometer as a noncontact sensing device and LDS Permanent Magnetic Shaker as an actuator. The obtained results show high susceptibility to noise and lesser accuracy in locating the debond zone, except the WTC and GFD. However, the WTC fails to provide good results for higher debond lengths, and the GFD shows prominent false alarms at the free edges of the plates. To circumvent these difficulties, two different modifications of the fractal dimension algorithm, namely the modified GFD (MGFD) and the GFD with GSM (GFD-GSM), have been proposed. Extensive numerical simulations are further carried out using commercially available finite element package ANSYS 14.0 in order to examine the experimental findings. In contrast to most previous work, the signal-to-noise ratio (SNR) in the experimental data has been appropriately quantified and noise of the same SNR level has been synthetically generated and applied on the modal data obtained from numerical simulations. Comprehensive studies for different debond locations and lengths suggests a similar trend as that obtained from the experimental investigations. Finally, a study on damage severity has been performed using the WTC and proposed modifications of the GFD. It is found that the proposed modifications of the fractal dimension perform outstandingly well in all circumstances, and can be used as an excellent tool for debond localization and quantification.