Shear Horizontal (SH) guided waves play a crucial role in non-destructive testing and structural health monitoring. Particularly, the higher-order modes of SH waves are sensitive to variations in material thickness, making it useful for detecting material thinning, while their group velocities can be conveniently studied through frequency control. This research proposed a novel omnidirectional high-order SH mode electromagnetic acoustic transducer (OHSHM-EMAT), composed of 14 axially polarized sector-shaped magnets with alternating magnetic field orientations, complemented by a spiral coil configuration. The transducer effectively excites and receives high-order SH1 guided waves in aluminum plates using a circumferentially periodic permanent magnet array. A 3D finite element model of the OHSHM-EMAT was constructed, successfully demonstrating the excitation of SH1 guided waves. Optimal configuration parameters for the magnets and coil were determined through optimization. To validate the accuracy of the simulation results, performance verification experiments were conducted on aluminum plates. Experimental results confirmed that the OHSHM-EMAT can generate omnidirectional SH1 mode guided waves in aluminum materials while effectively suppressing the generation of SH0 mode guided waves.

Acoustic velocity anisotropy in laser powder bed fusion (LPBF) components poses a significant barrier to highquality laser ultrasonic testing (LUT). This study presents an enhanced total focusing method (TFM) specifically designed to characterize submillimeter defects in anisotropic LPBF materials. Using specially fabricated specimens with artificial spherical pores, angle-dependent acoustic velocity corrections were implemented via polynomial and double Gaussian function fitting. A novel signal processing chain combining variational mode decomposition (VMD), multiscale principal component analysis (MSPCA), and outlier removal was introduced to stepwise eliminate strong scattering noise and Rayleigh wave interference. Crucially, a spatial-phase combined coherence imaging technique was developed, boosting the contrast ratio by over 40 dB and enabling the clear resolution of adjacent submillimeter pores. Experimental results confirm the superior spatial resolution and noise suppression capabilities. This work establishes a precise and efficient pathway for quality inspection in metal additive manufacturing.

Sensors play a vital role in nondestructive evaluation (NDE) and structural health monitoring (SHM), responsible for signal excitation and/or reception [...]

Conventional electromagnetic acoustic transducers (EMATs) are limited by their large dimensions for highprecision, high-resolution defect detection applications. This study presents an improved EMAT design utilizing a spatial vertical winding coil (SVWC), where both simulation models and experimental investigations were conducted to optimize the performance of the transducer. The experimental results demonstrate that incorporating a shielding layer in the SVWC configuration effectively eliminates interference from upper wires, suppresses acoustic side lobes, and reduces the coil's spatial height, achieving significant miniaturization. Compared to conventional planar coil EMATs with identical dimensions, the SVWC-EMAT exhibits enhanced transduction efficiency, showing 4.85-fold and 11-fold increases in signal amplitude on aluminum and 20# steel substrates, respectively. The implemented SVWC-EMAT achieves a minimum coil width of 1.0 mm, when integrated with synthetic aperture focusing imaging, successfully detects 1.0 mm diameter defects while resolving adjacent defects with 1.0 mm edge-to-edge spacing. These findings establish an important foundation for developing compact EMAT arrays and advancing high-precision electromagnetic acoustic imaging techniques.

To improve sensitivity to subsurface corrosion in conductive structures, this paper proposes a semi-ellipsoidal eddy current probe (SE-ECP) that integrates a curvilinear trapezoidal cross-section coil with a magnetic field sensor. In this study, an analytical formulation is derived. This is based on the extended truncated region eigenfunction expansion (ETREE). The aim is to obtain closed-form expressions of the magnetic vector potential, magnetic flux density and eddy current density. These are for Gaussian pulse-modulated eddy current (GPMEC) excitation in both the frequency and time domains. Numerical comparisons with finite-element modelling (FEM) demonstrate excellent agreement while showing that the ETREE solution is substantially more computationally efficient. Sensitivity analyses indicate that the SE-ECP produces a more spatially concentrated and deeper-penetrating eddy current distribution than a conventional pancake probe, yielding up to 36% improvement in sensitivity to material degradation and 31.7% improvement in sensitivity to wall-thinning under identical excitation. Experimental GPMEC imaging on aluminium specimens with engineered 3D localized subsurface corrosion qualitatively corroborates the sensitivity enhancement trends predicted by the theoretical and numerical models: SE-ECP images exhibit higher contrast and quantitatively larger differential peak responses, enabling more reliable detection and evaluation of subsurface corrosion. The amalgamation of analytical, numerical and experimental results demonstrates that the proposed SE-ECP significantly enhances GPMEC performance for quantitative subsurface corrosion assessment.

This paper presents an integrated non-destructive evaluation method for monitoring thermal aging in P91 steel by analyzing magneto-acoustic emission (MAE) signals through wavelet packet transform (WPT). Samples were thermally aged for 0–600 h at 780 °C and tested under controlled excitation conditions of 30 V and 30 Hz. The resulting MAE signals were processed using level-3 WPT decomposition to obtain energy distribution ratio (EDR

In industrial production, metal plates are widely used in various equipment and structures. Regular nondestructive testing of metal plates is essential for ensuring long-term operational safety and product quality. The electromagnetic acoustic transducer (EMAT) holds great potential due to its capability for non-contact detection, low surface roughness requirements, and ability to detect internal material defects. However, EMATs often face challenges such as low energy conversion efficiency, low signal-to-noise ratio (SNR) in echo signals, and limited spatial resolution. To address these challenges, this research proposes an encoded periodic permanent magnet shear horizontal wave EMAT (PPM SH EMAT) that incorporates pulse compression technology to simultaneously encode the excitation signal and the magnet configuration of the conventional PPM EMAT. Matched filtering and sidelobe suppression techniques are employed to enhance the SNR and spatial resolution of the transducer. Furthermore, by leveraging the magnetic field enhancement properties of Halbach array, the amplitude of defect signals is further boosted. Experimental results from defect detection of adjacent grooves in aluminum plates show that the proposed encoded Halbach shear horizontal wave EMAT demonstrates significant improvements in both SNR and spatial resolution.

To address the limitations of electromagnetic acoustic transducers (EMATs) in high-resolution defect detection, this article proposes a dense linear array EMAT based on spatial vertical winding coils (SVWCs) and an edge-constrained permanent-magnet array (EC-PMA). While ensuring high transduction efficiency for individual elements, this design constrains the element pitch to 1.5 mm, significantly increasing the number of elements within a finite aperture and enabling the development of a 32-element dense-array EMAT. The array performance was systematically characterized in terms of frequency response, lift-off attenuation, channel consistency, and interelement crosstalk. The results show that the elements exhibit broadband characteristics within the effective frequency range. Echo amplitudes decay exponentially with increasing lift-off, following a similar trend across different frequencies and materials. Channel consistency was evaluated using the relative pulse-echo sensitivity variation, and the results showed that 96.88% of the channels exhibited variations within $\pm$ 3 dB, indicating good uniformity. In addition, interelement crosstalk is closely related to both frequency and element spacing. It increases with frequency, reaches a maximum near 7 MHz, and decreases as the transmitter-receiver distance increases. By combining the developed array EMAT with a total focusing method (TFM) incorporating angle weighting factors, clear imaging of multiple side-drilled holes (SDHs) with diameters of 2.0, 1.0, and 0.5 mm was achieved in an 80-mm-thick P91 steel block. These results verify the feasibility and effectiveness of the proposed array EMAT for high-resolution defect imaging and provide a useful sensor foundation for advanced phased-array EMAT imaging.

Array-based non-destructive testing (NDT) has become a key advancement in modern NDT, offering large-area coverage, rapid imaging, and real-time data acquisition, significantly improving inspection efficiency and underpinning automated and intelligent inspection systems. Among array-based methods, electromagnetic acoustic transducers (EMATs) stand out as a non-contact method that eliminates the need for coupling agents, making them particularly well-suited for inspecting coated components and performing high-temperature in-line monitoring. Moreover, Array EMAT provide more intuitive, higher-resolution imaging, enhancing defect characterization and interpretation. Based on the full matrix capture of array EMAT, this study develops a fast frequency-domain signed coherence total focusing method. Derived from the rigorous solution of the wave equation, the frequency-domain algorithm offers clearer images compared to the time-domain delay-and-sum method. The introduction of the double-square-root vertical wavenumber significantly reduces the computational complexity of wavefield extrapolation. Meanwhile, the combination of frequency-domain zero-padding and time-domain interpolation effectively balances computational efficiency and spatial resolution. Gaussian filtering and the signed coherence factor are incorporated to suppress incoherent noise and mitigate the blind zones caused by electrical pulse crosstalk between array elements. Experimental results demonstrate that the array EMAT integrated with the advanced frequency-domain algorithm achieves an effective detection depth of 75 mm, with a single image computation time of only approximately 0.59 s. Moreover, it provides high-contrast and high-spatial-resolution detection results for multiple defects and adjacent defects. This research demonstrates the potential of array EMAT, combined with the proposed frequency-domain imaging algorithm, for future on-line non-destructive testing applications.

Ultrasonic guided waves (UGWs) have been widely used for the inspection of composite laminates. However, accurate signal analysis and damage localization depends on prior knowledge of ultrasonic guided waves dispersion information. To address this limitation, this study presents a sparse reconstruction framework for extracting ultrasonic guided waves dispersion information from limited measurement points. A non-convex sparse optimization model using the l1-2 norm sparsity constraint is formulated to enhance sparsity and enhance reconstruction reliability in the frequency wavenumber domain. Then, the proximal majorization-minimization algorithm is used to address the non-convexity and non-smoothness of the l1-2 norm. Furthermore, a sparse semismooth Newton solver is integrated into the proximal majorization-minimization (PMM) framework to ensure high solution accuracy and computational efficiency. The effectiveness of the proposed method is confirmed through both numerical simulation and experimental verification on a carbon fiber reinforced polymer (CFRP) laminate. The proposed method can reconstruct clear multimode dispersion curves across a broad frequency range of 20–1000 kHz using only 30 randomly selected measurement points. Robustness analysis over 20 independent trials confirms stable reconstruction performance. Compared with traditional two-dimensional fast Fourier transform, Lasso and l2-l1-2 method, the proposed method shows superior performance in reconstruction accuracy, frequency coverage, and energy distribution.

A lightweight long short-term memory (LSTM)-optimized time difference mapping (TDM) framework is presented for acoustic emission (AE) source localization in composite materials, establishing a systematic hardware-software co-design approach that evaluates the trade-off between localization precision and embedded resource constraints. TDM-derived features are integrated with an LSTM regression model to compensate for anisotropic wave propagation and multi-path dispersion inherent to composite structures. To address the rigid memory boundaries of microcontroller units (MCUs), an orthogonal experimental design is employed to optimize key hyperparameters, balancing deep model capacity against hardware limits. Model-compression techniques, including post-training int8 quantization, are subsequently applied to produce a lightweight TinyML model. Deployed with TensorFlow Lite for Microcontrollers on an ARM Cortex-M33 MCU, the quantized network occupies 78.9 kB of Flash and 6.1 kB of RAM. Empirical embedded evaluation demonstrates a viable balance between competing metrics, achieving a mean localization error of 10.89 mm, an inference latency of 2.8 ms, and a per-inference energy footprint of 143.9 μJ. These results validate the feasibility of embedding high-fidelity sequence-regression methods into resource-constrained edge hardware for decentralized structural health monitoring.

Accurate quantitative evaluation of surface crack depth is essential for the safety of metallic structures but remains challenging over a wide depth range. In this research, a Halbach magnet-based variable distance meander-line coil electromagnetic acoustic transducer (HBVD-EMAT) is used to generate wideband pulse compression surface waves for non-contact crack depth evaluation. The interaction mechanisms between wideband surface waves and surface cracks are investigated through theoretical analysis and experimental validation, focusing on frequency-dependent reflection, transmission, and diffraction-induced time-delay effects. Experimental results indicate that reflection and transmission coefficients are highly sensitive to shallow crack depths, enabling effective amplitude-based evaluation. For deeper cracks, severe attenuation of transmitted waves limits the reliability of amplitude characteristics. To overcome this limitation, time-of-flight differences between multiple transmitted wave packets are introduced, showing a strong linear relationship with crack depth and improving deep-crack evaluation accuracy. For small cracks, wideband signals are decomposed into multiple frequency components, and a neural network model is employed to capture the nonlinear relationship between crack depth and multi-frequency characteristics. The proposed multi-scale, multi-characteristic framework enables crack depth evaluation over a wide range, providing a practical EMAT-based solution for surface crack detection.

Terahertz defect detection in composites remains difficult when labelled samples are scarce, class distributions are imbalanced, and model predictions lack physical consistency. We propose a Physics-informed Constrained Inception-TCN-Attention Network (PI-ITCNANet) for small-sample terahertz signal analysis. The model supports defect classification, continuous depth estimation, and two- and three-dimensional defect imaging. The proposed model extracts multi-scale echo features using Inception modules. It uses dilated convolutions in the TCN module to capture long-range temporal dependencies among surface, defect, and bottom echoes. Channel and temporal attention mechanisms are further introduced to enhance defect-related responses. During training, multiple physical constraints are incorporated. These constraints encourage the predictions to satisfy both data supervision and the physical relationships of terahertz wave propagation. Experimental results show that PI-ITCNANet achieves Macro-F1 score of approximately 98.67%, depth MAE of 0.170 mm, and Dice coefficient of 0.957 for defect regions. The two- and three-dimensional reconstruction results further demonstrate that the proposed method can effectively recover defect locations, depth levels, and spatial distributions. These results indicate its potential as an intelligent recognition and imaging approach for terahertz nondestructive testing of composite materials.

To address the challenge of high-resolution detection of micro-defects in bimetallic materials, this paper proposes a laser ultrasonic testing method based on delay convolution and sum (DCAS). Unlike the conventional delay multiply and sum (DMAS), the DCAS algorithm replaces multiplication with convolution operations in the full matrix capture (FMC) data processing, effectively avoiding harmonic distortion and multi-mode signal interference while preserving spatial coherence. Furthermore, a sign function is incorporated to process delayed signals, and a grouping-based adjacent wave correlation subtraction technique is applied to enhance the signal-to-noise ratio (SNR). Experiments on aluminum-copper laminated specimens demonstrate that DCAS significantly outperforms the conventional delay-and-sum (DAS) and DMAS algorithms. It achieves maximum improvements of 82.8% in SNR and 514.4% in contrast-to-noise ratio (CNR), with a resolution enhancement of up to 50.0%. Notably, its computational efficiency is close to that of DAS, requiring only 68.9% of the computation time of DMAS, showcasing its strong potential for real-time, high-precision engineering applications.

The service life of thermal barrier coatings (TBCs) in aircraft engines is limited by high-temperature degradation and mechanical wear, necessitating non-destructive assessment to monitor their condition. To address the complexity of guided wave dispersion in coated structures, a zero group velocity (ZGV) Lamb wave-based approach is introduced for thickness measurement. This study proposes a method for characterizing thermal barrier coating specimens using ZGV Lamb modes, based on an all-laser ultrasonic excitation and detection system. Finite element simulations were performed to extract ZGV-related parameters, which were fitted to the coating thickness using least squares and multivariate regression analysis. Simple and multiple regression models were developed and validated through laser ultrasonic experiments. The results demonstrate a robust, linear relationship between selected ZGV parameters and coating thickness, allowing accurate estimation of layer thickness. The average relative errors of TBC and substrate thickness estimation are less than 7.71% and 1.55%, respectively. The combination of Lamb wave mode analysis and model-based regression offers a viable and non-contact method for the quantitative evaluation of ceramic coating thickness.

Glass fiber reinforced polymer (GFRP) composites are widely used in the aerospace field, but delamination defects are prone to occur during manufacturing and service. To address the challenges of low identification accuracy and difficulty in integrating depth information for delamination defects, this paper proposes an adaptive fusion imaging and quantitative characterization method based on reflective terahertz time-domain spectroscopy (THz-TDS). After filtering and denoising the time-domain signals, multiple imaging features are extracted using a sliding time window. An adaptive fusion algorithm based on the Sobel gradient is used to optimize the weight distribution of defects at different depths, achieving effective integration of multi-layer defect information. Local contrast is enhanced through background removal via polynomial fitting and morphological top-hat transformation. Gradient-weighted shrinkage algorithms and combined thresholding strategies are employed for defect segmentation and quantification. Experimental results demonstrated that variance features combined with adaptive fusion imaging could reveal multi-layer defects, with an average error of less than 7% for defects of different sizes. This method effectively suppressed background interference and significantly improved the identification sensitivity and localization accuracy of weak delamination defects within multilayer composites. Consequently, this paper provides a reliable non-destructive testing approach for the quality assessment and safety assurance of composite components.

This research aims to enhance the industrial viability of laser-ultrasonic full-matrix capture, which is often hindered by large data requirements and inefficient data acquisition. We present a novel framework for intelligent optimization imaging based on a multi-mode sparse array, with its configuration intelligently designed through Dynamic Multi-Swarm Particle Swarm Optimization (DMS-PSO). This method innovatively introduces the four ultrasonic propagation modes systematically into the sparse array optimization process, constructing a physics-driven multi-mode collaborative optimization model. The designed DMS-PSO algorithm synchronously optimizes the selection of arrays and the allocation of their operating modes, aiming to achieve an optimal balance between high sensitivity in the imaging region and uniform spatial coverage. Experimental results indicate that at a 20% sparsity rate, the multi-modal fusion method achieves a mean sensitivity of 4.08 with a standard deviation as low as 1.60, significantly outperforming single-mode optimization. Imaging of different types of submillimeter defects demonstrates that this method maintains excellent imaging performance across a sparsity range of 5% to 30%. Specifically, side drilling holes (SDH) defects exhibit imaging quality comparable to full-matrix results at 20% sparsity, while blind holes (BH) defects can also be effectively identified. Even at an extremely low sparsity rate of 5%, critical defect features remain identifiable. This study demonstrates the potential of the proposed multi-mode sparse array optimization method in enhancing the efficiency of laser ultrasonic testing. The results provide a promising technical basis for achieving online and rapid detection of micro-defects, although further validation on a broader range of materials and defect configurations is required to establish generalizability.

Ultrasonic guided waves (UGW) are the powerful tool for thin-walled structures to detect damage and monitor health conditions. Dispersion characteristics are the most important feature of guided waves and accurate dispersion curves are the application foundation of UGW. This study proposes a frequency-wavenumber sparse reconstruction method with generalized minimax concave penalty (GMCP) to reconstruct accurate guided waves dispersion curves for thin-walled structures using limited sensors. This method incorporates a generalized Huber-based non-convex penalty to promote sparsity in wavenumber domain reconstruction. An efficient forward-backward splitting (FBS) algorithm is derived to solve the optimization problem. To capture dispersion curves and gauge the quality of the reconstruction, a modes extraction method that combines an improved canny edge extraction algorithm with curve fitting is employed. Validations through simulations and experiments confirm the superiority of the proposed approach in high frequency range and high order modes.

Two-dimensional ultrasonic guided wavefield data carry rich waveguide structure information and are widely used for imaging and quantitative detection of structural damage. Focusing on the problem of high-volume data needs to be collected in the traditional damage imaging and quantitative detection method based on ultrasonic guided wavefield, this paper proposes a sparse multi-time wavenumber analysis method based on incomplete guided wavefield data. Firstly, the sparse sampling method of wavefield and the appropriate analysis dictionary are selected to construct the sensing matrix, and the compressed sensing equation is solved to reconstruct the snapshot of wavefield at multiple moments. Then, the continuous phase of the wavefield at multiple times is calculated, and the spatial phase gradient is calculated to obtain the wavenumber. After extracting the multi-time median wavenumber for each measurement point, the quantitative detection of damage is realized using the dispersion relationship between wavenumber and thickness. The method is first verified in the simulation, and then experimentally verified in an aluminum plate containing rectangular groove defects and a carbon fiber reinforced polymer plate with delamination defects. The simulation and experimental findings indicate that the proposed approach markedly decreases the quantity of measurement points while maintaining imaging quality and damage quantification accuracy.