Because the time reversal operator of Lamb waves varies with frequency in composite structures, the reconstructed signal deviates from the input signal even in undamaged cases. The damage index captures the discrepancy between the two signals without differentiating the effects of time reversal operator from those of damage. This results in the risk of false alarm. To solve this issue, a modified time reversal method (MTRM) is proposed. In this method, the frequency dependence of the time reversal operator is compensated by two steps. First, an amplitude modulation is placed on the input signal, which is related to the excitability, detectability, and attenuation of the Lamb wave mode. Second, the damage index is redefined to measure the deviation between the reconstructed signal and the modulated input signal. This could indicate the presence of damage with better performance. An experimental investigation is then conducted on a carbon fiber-reinforced polymer (CFRP) laminate to illustrate the effectiveness of the MTRM for identifying damage. The results show that the MTRM may provide a promising tool for health monitoring of composite structures.

Lamb wave based damage detection techniques have been widely used in composite structures. In particular, these techniques usually rely on reference signals, which are significantly influenced by the operational and environmental conditions. To solve this issue, this paper presents a baseline-free damage inspection method based on the reciprocity principle. If a localized nonlinear scatterer exists along the wave path, the reciprocity breaks down. Through estimating the loss of reciprocity, the delamination could be detected. A reciprocity index (RI), which compares the discrepancy between the signal received in transducer B when emitting from transducer A and the signal received in A when the same source is located in B, is established to quantitatively analyze the reciprocity. Experimental results show that the RI value of a damaged path is much higher than that of a healthy path. In addition, the effects of the parameters of excitation signal (i.e., central frequency and bandwidth) and the position of delamination on the RI value are discussed. Furthermore, a RI based probabilistic imaging algorithm is proposed for detecting delamination damage of composite plates without reference signals. Finally, the effectiveness of this baseline-free damage detection method is validated by an experimental example.

Accurate knowledge of dispersion relations is of vital importance for Lamb wave-based damage detection. In this paper, an experimental technique is proposed to obtain the Lamb waves phase velocity dispersion without prior knowledge of material properties. It requires measurements at only two neighboring positions, and can evaluate phase velocity dispersion over a wide frequency range. In that technique, the short-time chirp-Fourier transform is first applied to the Lamb wave signals to achieve a concentrated time–frequency representation. On this basis, ridge tracking is completed and the group delay (GD) of the interested modes is estimated accurately. Subsequently, the Vold–Kalman filter is introduced to extract these wave modes, and the instantaneous phase is estimated from each individual wave mode via Hilbert transform. Lastly, the phase velocity is obtained as a function of the propagation distance, GD and phase shift. The proposed technique is investigated in two different mediums: simulated signals as Lamb wave propagating in an aluminum plate and experimental signals as that in a carbon fiber-reinforced polymer laminate. The results demonstrate its effectiveness and efficiency.

The accurate information of Lamb wave packets, such as amplitude, time of flight, and waveform, are essential for damage detection and evaluation. Due to the dispersive, multimodal, and attenuative characteristics of Lamb waves propagating in viscoelastic media, the desired mode component is difficult to be extracted from response signals. To address this issue, a technique based on the viscoelastic propagation model of Lamb waves is proposed in this article. Different from the conventional frequency–wavenumber domain filtering, the proposed technique achieves mode extraction by estimating the parameters of the viscoelastic propagation model, that is, complex amplitude a_m( r, ω) and effective complex wavenumber ξ_m( ω). Benefiting from that, it enables to decompose and reconstruct the desired mode from spatial–temporal measurement data. First, the frequency-dependent effective complex wavenumbers are extracted using the matrix pencil method, where the real parts represent the dispersion and the imaginary parts represent the effective attenuation. After that, the complex amplitudes are obtained by solving linear least-square equations. Finally, the complex amplitude, effective complex wavenumber, and frequency range of the desired mode are picked out to reconstruct the mode component in the time domain. The finite element simulation and experimental results verify the effectiveness of our technique. Furthermore, the performance of the proposed technique in noisy circumstances is analyzed through synthetic signals, which shows a small reconstruction error and a good convergence.

This method based on Lamb waves shows great potential for long-range damage detection. Mode superposition resulting from multi-modal and dispersive characteristics makes signal interpretation and damage feature extraction difficult. Mode separation in the frequency–wavenumber ( f–k ) domain using a 1D sparse sensing array is a promising solution. However, due to the lack of prior knowledge about damage location, this method based on 1D linear measurement, for the mode extraction of arbitrary reflections caused by defects that are not in line with the sensor array, is restricted. In this paper, an improved compressed sensing method under the far-field assumption is established, which is beneficial to the reconstruction of reflections in the f – k domain. Hence, multiple components consisting of structure and damage features could be recovered via a limited number of measurements. Subsequently, a mode sweeping process based on theoretical dispersion curves has been designed for mode characterization and direction of arrival estimation. Moreover, 2D f–k filtering and inverse transforms are applied to the reconstructed f–k distribution in order to extract the purified mode of interest. As a result, overlapping waveforms can be separated and the direction of defects can be estimated. A uniform linear sensor array consisting of 16 laser excitations is finally employed for experimental investigations and the results demonstrate the efficiency of the proposed method.