Lamb wave-based damage assessment for composite laminates using a deep learning approach.

With the increasing utilization of composite materials due to their superior properties, the need for efficient structural health monitoring techniques rises rapidly to ensure the integrity and reliability of composite structures. Deep learning approaches have great potential applications for Lamb wave-based damage detection. However, it remains challenging to quantitatively detect and characterize damage such as delamination in multi-layered structures. These deep learning architectures still lack a certain degree of physical interpretability. In this study, a convolutional sparse coding-based UNet (CSCUNet) is proposed for ultrasonic Lamb wave-based damage assessment in composite laminates. A low-resolution image is generated using delay-and-sum algorithm based on Lamb waves acquired by transducer array. The encoder-decoder framework in the proposed CSCUNet enables the transformation of low-resolution input image to high-resolution damage image. In addition, the multi-layer convolutional sparse coding block is introduced into encoder of the CSCUNet to improve both performance and interpretability of the model. The proposed method is tested on both numerical and experimental data acquired on the surface of composite specimen. The results demonstrate its effectiveness in identifying the delamination location, size, and shape. The network has powerful feature extraction capability and enhanced interpretability, enabling high-resolution imaging and contour evaluation of composite material damage.

Iterative Pulse-Echo Tomography for Ultrasonic Image Correction.

Acoustic aberration, caused by the uneven distribution of tissue speed-of-sound (SoS), significantly reduces the quality of ultrasound imaging. An important approach to mitigate this issue is imaging correction based on local SoS estimation. Computed ultrasound tomography in echo mode (CUTE) is an SoS estimation method that utilizes phase-shift information from ultrasound pulse-echo signals, offering both practical utility and computational efficiency. However, the traditional single-pass CUTE often suffers from poor accuracy and robustness. In this paper, an advanced approach known as iterative CUTE is introduced, which refines SoS estimates through iterative correction of errors and noise, addressing the limitations of traditional single-pass methods. It is argued that traditional precision indicators like root mean square error (RMSE) fall short in adequately reflecting the quality of SoS estimates for imaging correction, and coherence factor (CF) is proposed as a more indicative metric. Performance validation of the iterative CUTE algorithm was conducted using a simulation and agar phantom experiment. The results indicated that the iterative CUTE approach surpasses the single-pass approach, enhancing the average CF for SoS estimates by up to 18.2%. In phantom experiments, imaging corrected with SoS estimates from iterative CUTE reduced the Array Performance Index (API) by up to 40% compared to traditional methods.

Energy focusing of broadband Lamb wave by designing excitation waveforms and elastic metamaterials.

In the field of structure health monitoring (SHM), the use of Lamb wave to locate damage is a common method. Energy focusing is beneficial for damage localization because of higher SNR and higher resolution. Optimization design of elastic metamaterials is promising for energy focusing based on speed modulation. However, current design scheme is effective only for narrowband Lamb waves. Compared to narrowband waves, broadband Lamb waves with a larger frequency range carry richer structural information. In this study, an energy focusing method based on broadband Lamb waves by simultaneous designing excitation waveforms and elastic metamaterials is proposed. Firstly, COMSOL finite element simulation software is used to calculate the relationship between the metamaterial structure and the excitation wave. Subsequently, the metamaterial structure and the excitation signal form are designed according to the relationship. Finally, the metamaterial structure is bonded with the aluminum plate using 3D printing PA2200 nylon to verify the effectiveness of the method.

Quantitative evaluation of crack based on the sparse decomposition of array Lamb wave propagation.

Crack damage is one of the significant factors that may accumulate at the stress concentration area of engineering structures and cause catastrophic accidents. In this paper, we proposed a novel approach to identify the crack location and size by exploiting the reflections and diffractions of Lamb waves. The interaction mechanism between the crack and Lamb wave has been analyzed thoroughly, our analysis of the interaction between the crack and Lamb wave revealed that both reflections and diffractions carry valuable damage information that reflects the size and orientation of the crack. As the interaction coefficients between these two components and the crack are different, there are supposed to exist differences between them in the amplitude value. We implemented a threshold to classify the signals received from all paths into two groups: reflections and diffractions. Then we constructed an overcomplete dictionary of waveforms corresponding to different propagation distances to extract the damage information. Using sparse decomposition, the received signals were mapped to their corresponding propagation distances without the use of baseline signals. The diffractions allow us to determine the crack's tip points, while the reflections provide information about the edge points. The kinked crack's size and orientation were visualized based on the time-domain signals acquired in our experiment. We provided a comprehensive description of our algorithm and verified it through numerical simulation and experimental data. Our results show high agreement with actual cracks, demonstrating the efficacy of our proposed method.

Damage characterization using CNN and SAE of broadband Lamb waves.

The method based on Lamb wave shows great potential for structural health monitoring (SHM) and nondestructive testing (NDT). Deep learning algorithms including convolutional neural networks (CNN) and stack autoencoder (SAE) are promising to extract features from Lamb wave signals that can be linked with damage for subsequent localization and quantification. Generally, narrowband Lamb wave with purified mode and suppressed dispersion is used because of clear relationship model between damage features and recorded signals. However, model performance is limited because contained damage information of narrowband Lamb wave is inadequate. To overcome this limitation, a broadband Lamb wave deep learning algorithm is proposed for damage localization and quantification. Compared with narrowband, broadband Lamb wave generated at a large frequency range contains richer information of structural damage. In this study, different mode selections, different signal processing methods and different deep learning algorithms are applied to extract damage features from different perspectives, and fusion of all extraction results facilitates the full utilization of rich broadband information. An experiment is given to demonstrate the effectiveness and high-accuracy of proposed method.

Improved spectrum method for impact damage characterization in the composite beam using Lamb waves.

Lamb wave spectral methods are one candidate to characterize invisible damages in composite structures. Unfortunately, multiple reflections resulting from geometric boundaries could distort the Lamb wave spectrum, which may cover the important signatures concerning structural integrity. To eliminate spectral interference, a cepstrum based filtering method is proposed to separate various reflection features. In particular, the fundamental spectrum contributing to structural integrity can be extracted smoothly by removing harmonic fluctuations regarding wave reflections with an optimized filter. Subsequently, to establish the quantitative relationships between fundamental spectrum features and the impact energies, damage indices involving spectrum energy mean and median frequency shift are introduced to quantify the change of spectrum intensity and distribution respectively, which shows good performance on impact damage characterization. Finally, the experiment was carried out on a T300 composite beam, in which strong reflections come after direct waves. Meanwhile, the severity of impact damage was simulated by the free droppings of a steel ball with different heights. The results show the effectiveness of the proposed method for characterizing impact damages with improved sensitivity.

Local Wavenumber Method for Delamination Characterization in Composites With Sparse Representation of Lamb Waves.

This article presents a quantitative Lamb wave detection method for delamination characterization in composite laminates using local wavenumber features. In contrast to the conventional Fourier transform-based methods, the improved sparse reconstruction method is efficient and able to evaluate the spatial wavenumbers of Lamb waves with limited measurements. To improve the feasibility of the sparse reconstruction method, the analytical solution of the local wavenumbers to the compressed sensing (CS) formulation with considering structural discontinuity is firstly investigated. The estimated wavenumber values for the spatial window located on healthy and damaged regions simultaneously are governed by a nonlinear optimization function. Benefiting from revealing the evolution of local wavenumber slopes, a delamination characterization method is proposed to accurately determine the damage location and depth by wavenumber banalization. Subsequently, the performance of the proposed method on local wavenumber estimation and damage quantification is verified by the simulation data. The parameters are discussed in order to improve the algorithm stability. Finally, the experimental investigation was conducted on a quasi-isotropic carbon fiber reinforced polymer (CFRP) laminate with an induced delamination. Lamb wave was generated and scanned by the Nd:YAG laser spot with spatial intervals of 2 mm. The results verify the sparse reconstruction method for delamination characterization, which shows value of reducing labor cost and testing time.

Amplitude modified sparse imaging for damage detection in quasi-isotropic composite laminates using non-contact laser induced Lamb waves.

Composite structure is increasingly used in civil and aerospace applications due to its high mechanical performance. Lamb wave based sparse reconstruction imaging for damage localization is promising for structural health monitoring (SHM) and nondestructive evaluation (NDE) by using few measurements. However, this dictionary based method requires accurate atoms to represent Lamb wave propagating features in structure very well. Besides dispersion, signal changes caused by amplitude modulation should be considered for waveform distortion when constructing the dictionary for sparse imaging method. In this paper, a non-contact laser is used for Lamb wave excitation which exhibits a strong amplitude modulation in low frequency. Additionally, the strong attenuation resulting from material damping would also presents a distance-dependent amplitude modulation. To reconstruct an amplitude model of Lamb wave, the decomposition method of system response and attenuation is proposed. Then, the influence of amplitude modulation on signal representation is analyzed, which shows the restriction of dictionary without considering amplitude modulation. On this basis, the amplitude considered dictionary is built together with the phase considered dictionary for sparse imaging in terms of damage detection. Furthermore, according to Lamb wave reflection model, the solution for sparse reconstruction imaging is given. Finally, the performance of sparse imaging method is discussed by experimental investigation with different parameters. The results show the efficiency of the proposed method with improved imaging performance and give comparisons for better parameter choice.

Minimum variance imaging based on correlation analysis of Lamb wave signals.

In Lamb wave imaging, MVDR (minimum variance distortionless response) is a promising approach for the detection and monitoring of large areas with sparse transducer network. Previous studies in MVDR use signal amplitude as the input damage feature, and the imaging performance is closely related to the evaluation accuracy of the scattering characteristic. However, scattering characteristic is highly dependent on damage parameters (e.g. type, orientation and size), which are unknown beforehand. The evaluation error can degrade imaging performance severely. In this study, a more reliable damage feature, LSCC (local signal correlation coefficient), is established to replace signal amplitude. In comparison with signal amplitude, one attractive feature of LSCC is its independence of damage parameters. Therefore, LSCC model in the transducer network could be accurately evaluated, the imaging performance is improved subsequently. Both theoretical analysis and experimental investigation are given to validate the effectiveness of the LSCC-based MVDR algorithm in improving imaging performance.

Excitation Waveform Design for Lamb Wave Pulse Compression.

Most ultrasonic guided wave methods focus on tone burst excitation to reduce the effect of dispersion so as to facilitate signal interpretation. However, the resolution of the output cannot attain a very high value because time duration of the excitation waveform cannot be very small. To overcome this limitation, a pulse compression technique is introduced to Lamb wave propagation to achieve a δ-like correlation so as to obtain a high resolution for inspection. Ideal δ-like correlation is impossible as only a finite frequency bandwidth can propagate. The primary purpose of this paper is to design a proper excitation waveform for Lamb wave pulse compression, which shortens the correlation as close as possible to a δ function. To achieve this purpose, the performance of some typical signals is discussed in pulse compression, which include linear chirp (L-Chirp) signal, nonlinear chirp (NL-Chirp) signal, Barker code (BC), and Golay complementary code (GCC). In addition, how the excitation frequency range influences inspection resolution is investigated. A strategy for the frequency range determination is established subsequently. Finally, an experiment is carried out on an aluminum plate where these typical signals are used as excitations at different frequency ranges. The quantitative comparisons of the pulse compression responses validate the theoretical findings. By utilizing the experimental data, the improvement of pulse compression in resolution compared with tone burst excitation is also validated, and the robustness of the waveform design method to inaccuracies in the dispersion compensation is discussed as well.