Ultrasonic Guided Waves Research Articles

Probabilistic diagnostic imaging method based on the difference between the arrival moments of ultrasonic guided wave signals

Ultrasonic guided wave (UGW), using the probabilistic diagnostic imaging (PDI) method is widely used in structural health monitoring. The PDI method, which incorporates signal time-of-flight (TOF), has significant advantages in high-precision localization and multi-damage recognition. However, the TOF acquisition process introduces errors. This paper proposes a modified PDI method based on the difference between the arrival moments of guided wave (GW) signals. This method eliminates the need to acquire excitation to obtain TOF, reducing the requirements for GW detector, and minimizing the impact of TOF errors on damage localization. An experimental system for composite laminates was constructed, and the experimental results showed that the method had an average localization error of 4.73 mm for single damages. The average localization error for both multiple damages and damage at edge locations was less than 20.0 mm. The localization error is satisfactory with reference to the size of the monitored area and damages.

Optimising structural health monitoring systems: A 2D numerical investigation on impedance matching for FML with integrated sensors

AbstractFibre metal laminates (FML) represent an innovative class of advanced composite materials that integrate the mechanical properties of both metals and fibre‐reinforced composites (FRP). Combining the strength and ductility of metals with the lightweight and high stiffness of FRP and FMLs have emerged as new material compositions for applications in chemical, nuclear, automobile, and aerospace engineering disciplines. Structural health monitoring (SHM) using guided ultrasonic waves (GUW) is the state‐of‐the‐art for non‐destructive testing of thin‐walled structures. When applied to FML, SHM plays a crucial role in monitoring the integrity over time and detecting potential damage such as delamination, fibre breakage, or other structural anomalies. In SHM with GUW, a wave‐field is emitted by actuators. This wave‐field can be affected by damage in the structure, thereby changing its propagation characteristics. Sensors monitor the interaction between damage and GUW, which can be utilized to locate and classify the damage and ascertain the overall health state of the structure. In this study, an advanced integration of measurement hardware, that is, sensors and actuators, within the laminate structure is investigated. Sensor integration into FML allows for improved and more sophisticated monitoring capabilities in comparison to measuring techniques like laser vibrometers, which are limited to measuring displacements on the surface of the structure. However, the integration of sensors and actuators yields the technical difficulty of distorting the wave‐fields and may result in an over‐ or underestimation of the damage. Similar to damage, the distortion of the wave‐field is caused by the changes in acoustic impedance resulting from different material properties. In a previous study, incorporating a functionally graded artificial interphase through acoustic impedance matching between the sensor and host material showed notable and significant outcomes. The current contribution extends the prior graded artificial interphase for an isotropic homogeneous material to an FML structure. This paper presents a comprehensive numerical simulation study on a two‐dimensional model of FML with integrated sensors. The interphases are designed based on impedance matching, which improves signal transmission and reduces disturbing reflections. The conducted investigations hold for several interphase configurations for a wide frequency range. The optimised integration of sensors demonstrates promising results for enhancing the reliability and accuracy of SHM systems. This research serves as a foundation for further experimental validation and the development of advanced sensor‐integrated FML structures with improved monitoring capabilities.

Ultrasonic guided wave damage localization method for composite fan blades based on damage-scattered wave difference

Abstract Ultrasonic guided wave (UGW) has a wide monitoring range and high accuracy, holding promise for monitoring damage in engine large-area composite fan blades. However, the multi-curvature characteristics of engine composite fan blades with their anisotropic material properties make damage localization impossible with conventional UGW monitoring methods. In order to realize the UGW damage monitoring of the blade, this paper proposes a damage localization method based on damage-scattered wave difference. The method utilizes irregular sensing arrays to achieve localization of damage in multi-curvature composite blades with a small amount of data. First, the difference between the mutual excitation in a pair of sensors and the damage-scattered waves captured at reception was analyzed. It is concluded that the closer the damage is to the received sensor, the greater the damage index (DI). Next, a DI ratio of the mutually excited and received signals is computed for each sensor pair. This ratio is utilized to make a vertical line on the propagation path, which is identified as the damage likelihood line (DLL). Finally, a DLL corresponding to the three largest DIs was selected, and their intersections were used for damage localization. A time-domain truncated signal processing method is proposed to enable DI to more accurately represent the effects of damage, and improve the localization accuracy of the method. An experiment on damage localization was carried out on a homemade composite fan blade, where the damage was tested at various locations and sizes. The results show that the damage localization on the blade is good and 3mm tiny damage localization is achieved.

Ultrasonic guided Wave Signal Denoising Method Based on one-dimensional convolution and Full connection Model Fusion Denoising autoencoder

Abstract Ultrasonic guided wave (UGW) detection is widely used in pipeline monitoring but faces challenges from weak flaw echo signals within the detection data, making weak UGW signals difficult to recognize. It is essential to denoise the UGW detection signals to identify weak echo signals. This paper proposes an improved denoising autoencoder (DAE) based on the fusion of one-dimensional convolution neural network (1DCNN) and full connection (FC). The model expands the amount of training data by adding noise in batches and uses 1DCNN to enhance the ability of extracting UGW signal features. The model was validated using Several numerical simulation signals. Numerical simulation results show that the signal-to-noise ratio (SNR) of the UGW signals can be improved from -20 dB to 8 dB; it has a strong improved SNR, and the mean square error is greatly reduced while maintaining the original phase almost unchanged. The improved DAE method has significant advantages in denoising effect, and it can effectively reduce the noise of the UGW detection signal and realize the identification of small defects of the simulation pipeline.

Ultrasonic guided wave-based probabilistic diagnostic imaging method with Single-Path-Scattering sparse reconstruction for Multi-Damage detection in composite structures

Plate-like carbon fiber composite structures are essential for equipment such as spacecraft and rail vehicles. These structures are exposed to long-term cyclic loads and harsh environmental conditions, leading to damage of unknown location and quantity. Ultrasonic guided wave (UGW) detection technology shows promise for evaluating the structural performance of composite structures due to its advantages of high precision, high speed, and real-time capabilities. However, multi-damage scattering sources interfere with and superimpose the received signals when multiple damages coincide. This interference can lead to artifacts in probability imaging methods that rely on features such as amplitude and time of flight (TOF). This paper proposes a single-path-scattering sparse reconstruction-based probabilistic diagnostic imaging (SSR-PDI) method. Firstly, a Lamb dictionary with time shift is constructed. Each actuator-receiver path is treated as a target signal, enabling the approximate separation and extraction of multiple scattering waves from different damages. It mitigates the “curse of dimensionality” associated with the sparse reconstruction (SR) method. Secondly, a new ring-shaped damage probability distribution is proposed based on the sparse representation matrix’s time mapping and amplitude weighting. The experimental results demonstrate that the SSR-PDI method effectively mitigates issues related to multi-damage detection, specifically false negatives/positives and localization errors arising from the superposition of scattering sources.

A Multi-Strategy Hybrid Sparse Reconstruction Method Based on Spatial-Temporal Sparse Wave Number Analysis for Enhancing Pipe Ultrasonic-Guided Wave Anomaly Imaging.

Ultrasonic-guided waves (UGWs) in defective pipes are subject to severe coherent noise caused by imperfect detection conditions, mode conversion, and intrinsic characteristics (dispersion and multiple modes), inducing the limited performance of anomaly imaging. To achieve the high resolution and accuracy of anomaly imaging, a multi-strategy hybrid sparse reconstruction (MHSR) method based on spatial-temporal sparse wavenumber analysis (ST-SWA) is proposed. MHSR leverages the capability of ST-SWA to extract the wavenumber dispersion curves, thereby providing a more refined and precise search space for MHSR. Furthermore, it mitigates the impact of coherent noise by conducting dispersion compensation on the reconstructed signal. The sparse compensated signals through MHSR are employed for sparse reconstruction imaging. To validate the efficacy of the proposed method, UGW testing is performed on the defective steel pipe, and the results demonstrate the significant enhancement of anomaly imaging in defect resolution and positioning accuracy. The lowest estimated errors for axial and circumferential defect positions are 10 mm and 4 mm, respectively.

Guided wave-driven machine learning for damage classification with limited dataset in aluminum panel

The goal of this research study is to classify damages in an aluminum panel using ultrasonic-guided waves (UGWs). The study is carried out on the premise that only a limited dataset, consisting of only two receivers, is available. In the experimental phase, three distinct parts, that is, stainless steel masses, were located on the surface of the panel, assumed to be mapped into four triangular regions, to experimentally simulate the effects induced by circular artificial damage of varied sizes. UGW-driven damage classification was thus performed via machine learning. A data synthesis technique called conditional generative adversarial network (cGAN) was used to generate 2000 samples (signals) for each of the 12 different malfunction scenarios (classes), based on the combination of damage region and size. Damage features were extracted using the wavelet time scattering technique, while the classification task was performed via a long short-term memory network with tuned hyperparameters. A validation procedure was executed to ensure that the classification network was not overfitted. Furthermore, two testing stages were performed to evaluate the designed framework. The designed framework was thus tested on the observations excluded from the training and validation phases, as well as on 10 additional novel experimental samples. The high accuracy and F1-score values of the validation and the two testing stages—97.7%, 98.5%, and 96.6%, respectively—attest to the generality of this methodology, which is neither underfitted nor overfitted. To further assess the robustness of the methodology, the cGAN was tested on a new sensor placement, demonstrating a consistent performance (95.3%) and confirming the framework applicability under different sensor configurations, showcasing potential extension to other classification problems featuring a limited dataset.

An interpretable TFAFI-1DCNN-LSTM framework for UGW-based pre-stress identification of steel strands

Steel strands serve as the key load-bearing components of pre-stressed bridges, yet the identification of effective pre-stress for steel strands is a challenging task. In this study, a time and frequency adaptive fusion input-one dimensional convolutional neural network-long short-term memory (TFAFI-1DCNN-LSTM) framework was proposed for ultrasonic guided wave (UGW)-based effective pre-stress identification of steel strands. In this framework, the time and frequency domain signals of UGW were input into the network as two parallel branches, CNN and LSTM were utilized to extract spatial and serial-related features, and a trainable weight coefficient was introduced for adaptive fusion of time and frequency domain features. Firstly, ablation studies were conducted to investigate the influence of each key component in the proposed framework on the prediction results. Secondly, deep learning (DL) models of Res Net, Dense Net and Inception Net were introduced as comparisons, and the prediction results based on TFAFI-1DCNN-LSTM were compared with those based on the above DL models. Thirdly, the noise-robustness and performance under a few trainable samples of TFAFI-1DCNN-LSTM were also investigated. Finally, the outputs of various key layers of the proposed framework were subjected to dimensionality reduction and visualization to provide an intuitive demonstration of the feature extraction process, and the inherent mechanisms of the proposed framework were interpreted using local interpretable model-agnostic explanations (LIME). Results show that the architecture of TFAFI-1DCNN-LSTM is reasonably designed and all key components are necessary. The weights of features in time and frequency domain branches are rationally assigned due to the addition of an adaptive weight coefficient. Compared with other DL models, the proposed TFAFI-1DCNN-LSTM exhibits higher pre-stress identification accuracy, excellent noise-robustness, and superior performance under only a few trainable samples. The correlation between features extracted by the network and the pre-stress labels significantly increases with the deepening of the network, and the correlation between input UGW signals and predicted pre-stress values is effectively interpreted utilizing LIME, proving that the architecture of the proposed framework is reasonable and effective, the predicted results are reliable and credible.

Damage detection method for square steel tube based on CS-NME algorithm via ultrasonic guided waves

The utilization of ultrasonic guided wave (UGW) technology has garnered considerable attention in non-destructive testing field, particularly for steel components. However, for square steel tubes with noncircular cross sections, this method encounters severer frequency dispersion, which hampers the accuracy of damage identification. To address this issue, the approach utilizing the normal mode expansion (NME) method combined with cuckoo search (CS) algorithm to develop a layout strategy for ultrasonic transducer position and length is proposed, which enables single mode excitation of UGW while reducing the interference from redundant clutter signals, thereby enhancing the accuracy of damage detection. The efficacy of the proposed approach is evaluated via numerical simulations and laboratory experiments. The results show that this method can significantly reduce the interference of other modes by optimizing the location and length of ultrasonic transducers, and successfully detect the hole and crack damage on the square steel tube.

Impact of the Metal Volume Fraction on the Propagation of Guided Ultrasonic Waves in Fiber Metal Laminates

This study investigates the impact of metal volume fraction (MVF) on the propagation of guided ultrasonic waves (GUW) in fiber metal laminates (FML). GUW interactions with various metal and prepreg layers are influenced by changes in the FML composition. A crucial parameter for FML characterization is the metal volume fraction (MVF), representing the volume of metal to prepreg. This research assumes that variations in MVF affect GUW propagation in FMLs. The study analyzes the influence of MVF on GUW propagation in FML, specifically comparing GUW propagation in GLARE plates with nearly identical geometry but different MVF and layer structures. Initially, simulations were conducted to predict expected changes in dispersion properties. To validate these predictions, pitch-catch measurements were performed on the panels, measuring three transducer paths for each panel to enhance measurement significance. An intensive study of wave modes was undertaken to investigate the influence of MVF on mode velocities. The simulation results were validated through experiments, highlighting the significant impact of MVF on GUW propagation in FML. Interestingly, in FMLs with varying MVFs and layer structures, the structural conditions of the layers (such as fiber orientation) demonstrated a non-negligible influence on GUW propagation, surpassing the influence of MVF alone.

Semi-analytical simulation of ultrasound wave propagation in large plate-like structures

The ultrasonic guided wave (UGW) has gained considerable popularity among NDE methods for damage detection and structural health monitoring (SHM) due to its high frequency and low attenuation. Lamb waves are reflected or diffracted by structure boundaries, discontinuities, and damages. It is possible to determine a structure's health, defects, and locations based on its UGW propagation characteristics. In this study, we propose an efficient modeling method for UGW propagation through damaged and scattering plates by combining the Modal Pencil method with Wave Finite Element (WFE) and Hybrid FE/WFE methods and verifying the results with a high-precision elastic finite element model (FEM). With this approach, the transducer-excited ultrasonic field can be modeled, as well as its steady-state and transient propagation, scattering, and reflections at defects and boundaries. Long-range propagation configurations can be efficiently addressed by this method because its computation efficiency is independent of propagation path lengths. As part of the case study, we study the propagation of Lamb waves through a joint metal thin-walled structure with embedded damage. In the case study, different sensor locations are used to validate the received UGW signals quantitatively. Our findings will contribute to the development of a digital twin for monitoring structural health by providing an efficient and robust approach to modeling ultrasound propagation in damaged and large plate-like structures.

Data-driven Parameterizable Generative Adversarial Networks for Synthetic Data Augmentation of Guided Ultrasonic Wave Sensor Signals

Detecting and characterizing hidden damages in composite materials like Fibre-Metal Laminates (FML) remains a challenge. Guided Ultrasonic Waves (GUW) or X-ray imaging are commonly used to detect these damages, but their interpretation remains limited. Data-driven predictor models can detect damages in structures using GUW time-dependent signals, but the training data lacks variance, quality, and sufficient coverage of the hyper-parameter space. Synthetic data augmentation is often the only solution to create robust and generalized damage predictor models. Synthetic sensor data can be generated using model-based, model-assisted, or model-free methods. However, computed GUW signals show poor reality conformance due to too much constraints and simplifications. Recent developments in Generative Adversarial (Neural) Networks (GAN), commonly driven by a random generation process [1], include deterministic style vectors to generate specific signal data, determining constraints such as damage size, position, transducer positions, material and environmental properties. These new architectures aim to reduce the impact of environmental changes on data-driven damage predictor models by using parameterizable synthetic data generation. We will elaborate these new architectures and discuss the possibilities and challenges of using such GAN models for data generation of US signals in GUW-based SHM applications, as originally proposed in [2]. We will have a focus on low-quality real measuring data used for training and composite materials (e.g., FML) with hidden damages, and show some preliminary results. [1] Karras et al. "A style-based generator architecture for generative adversarial networks." Proceedings of the IEEE/CVF 2019 [2] Virupakshappa et al. "Using Generative Adversarial Networks to Generate Ultrasonic Signals." 2020 IEEE International Ultrasonics Symposium (IUS). IEEE, 2020

Reconstruction of direct waves between passive flexible sensors using diffuse ultrasonic waves

The sensing system that enables generation and sensing of guided ultrasonic waves (GUWs) serves as a pivotal cornerstone for the structural health monitoring based on GUWs. Traditional sensing systems have been constrained by inherent limitations including substantial mass addition and limited flexibility, and this leads to an increasing research efforts for the emerging nano-composite sensors. Despite the superior properties of lightweight and flexibility, the nano-composite sensors is incapable of active wave excitation, and thus waves propagating directly between sensors cannot be obtained. To tackle this deficiency, an approach grounded in representation theorem to reconstruct signals using diffuse ultrasonic waves is proposed. In this approach, the waves propagating between two passive sensors can be approximated through the cross-correlation of wave signals in a diffused status. On this basis, the passive sensors can be virtually converted to active actuators, enabling the acquisition of waves virtually excited and captured by a pair of passive sensors. In experimental validations, the wave signals between passive sensors are retrieved and the damage close to the sensing path is detected. The proposed method enhances signal acquisition capability of sensing networks based on flexible passive sensors and opens avenues for the monitoring of previously inaccessible blind areas within complex structures.

A Study on XANNs for Analyzing Failures in Guided Ultrasonic Wave-based Damage Localization

In the field of Structural Health Monitoring (SHM) using Guided Ultrasonic Waves (GUW), machine learning (ML) approaches for data analysis are becoming increasingly popular. Especially, deep learning with artificial neural networks (ANN) has proven effective due to its ability to process a large amount of data. ANN are commonly employed in damage localization tasks, where, with appropriate training, they deliver good results. However, a significant challenge persists in the application of ANN, namely, their "black-box" nature, which hinders the investigation of decision processes within the ANN and makes it difficult to understand why an ANN model fails under certain inputs. In the present work, an already introduced explainable artificial neural network (XANN) approach will be utilized to investigate when and why a trained XANN model fails with given inputs. The Open Guided Wave (GUW) platform's data is employed to train and test a XANN model designed to predict the location of artificial damage. Subsequently, individual inputs are modified (simulating malfunctions in specific sensor paths), and the impact on decision processes within the XANN architecture is examined. The aim of this study is to investigate and enhance the resilience and explainability of ANN approaches in the SHM domain.

The usage of interference effects in single Guided Ultrasonic Wave time signals for damage identification

Guided Ultrasonic Waves (GUW) interact with structural inhomogenities in multiple ways, e.g. mode conversions, mode scattering, oscillation phase shifts and reflections. These effects do occur at specimen edges and at damage on and in the specimen’s area. These effects at structural boundaries make the signal interpretation more complex. On the other hand they can lead to additional information in recorded time signals when caused by damage. This study aims at the usage of interference effects in single time signals caused by damage in the structure. This is investigated changing the location of the measurements points in relation to the damage position. Two cases are taken into account: once the damage lies in the direct measurement patch between actuator and measurement point and once the damage lies outside the measurement path and its effect on the recorded time signal is investigated. In the experimental setup, a rectangular Aluminum plate is used. Its dimension is chosen to make sure that edge reflections do not superimpose with the regarded wave packages within the time signals. A Laser-Doppler-Vibrometer allows the non-contact recording of several measurement points and is not wavelength-dependent as e.g. PZT-transducers. This allows the measurement to be unaffected by the transfer behavior of the transducers. The relationship between the specific mode wavelength reflected by the damage and the measurement path length provides information about the expected level of interference at the considered measurement point. The anticipation includes the indirect identification of damage off the direct measurement path through the observed interference effects, as well as variations in interference caused by the shift of the measurement point locations. The amplitude and phase changes serve as a measure of the interference and are obtained using the Hilbert transform.

Reducing Artefacts in FRF-based Damage Detection Using Novel Mode Extraction Approach for Guided Ultrasonic Waves

Damage in thin-walled structures can be detected by guided ultrasonic wave (GUW) based structural health monitoring systems. The application of phased arrays enables the scanning of large-scale structures from a single position. However, physical wave focusing requires a lot of effort. This can be significantly reduced if frequency response functions (FRFs) are used. They enable the calculation of virtual response signals for virtual focusing on any position. Although this technique offers an energy-efficient and fast possibility of damage detection, it has the disadvantage of artefacts that occur due to the multimodal nature of GUW, as damage detection is generally designed for single-mode signals. These artefacts are often reduced by mode-selective excitation/sensing or by subtracting a baseline measurement. This work presents a concept to combine an existing FRF-based damage detection algorithm and a previously presented method for mode extraction. It enables the extraction of GUW mode components from broadband, temporally sampled, single-input single-output sensor data during signal processing on the basis of the respective dispersion relations. The aim is to reduce artefacts without using mode-selective excitation/sensing or baseline measurements. A finite element simulation of GUW propagation in an isotropic structure is used to demonstrate the advantages and limitations of this approach. The simulations include multimodal and single mode evaluation to point out the added value of performing mode extraction prior to damage detection. It is supposed that the successful extraction of the mode components from the temporally sampled data results in a decreasing amplitude of the occurring artefacts compared to the multimodal case, while the case of mode-selective excitation apparently results in no artefacts. The presented concept of adding mode extraction to damage detection algorithms can lead to an increase in performance of FRF-based phased array systems. Artefacts that would lead to false detection of damage are reduced inherently during signal processing which eliminates the need for mode-selective excitation/sensing or baseline measurements.

Convolutional Autoencoders for Signal Reconstruction and their Application to Damage Signature Extraction

The utilization of Machine Learning (ML) techniques for Structural Health Monitoring (SHM) has increased in the last years. Using ML techniques for signal reconstruction has been explored in the literature of this field. In this work, a database of Ultrasonic Guided Waves (UGW), propagated through a thermoplastic composite plate, and generated and acquired with piezoelectric transducers (PZT), has been preprocessed, organized, and analyzed to reconstruct signals by using Convolutional Autoencoders (cAE) neural networks. This special type of autoencoders (AE) is based on the utilization of convolutional stages, which are formed by deep-learning layers that are especially useful for their utilization on discretized time-series signals, due to their ability to convolve different sections of the signal, thus extracting different features which are automatically learned by the network. The cAE, which have a bottleneck section in their middle part (between the encoder and the decoder sections), have the ability to reduce the dimensionality of the input data, thus allowing the extraction of a minimum number of relevant features of the signals. Summarizing, this paper presents a neural network designed as a cAE which is able to reconstruct data from a reduced number of features, reaching a correlation value between the real and the artificial data higher than 98%.

Non-Contact Laser Doppler Vibrometer for Characterization of Guided Waves in Timber Structures with Varying Moisture Content and Grain Orientations

Timber structures have been widely used in critical engineering applications throughout history. The need for Non-Destructive Testing of timbers has arisen, and ultrasonic-guided waves (GWs) have proved their efficacy in studying structural integrity. The effect of the grain direction in orthotropic timber structures on the propagation characteristics of GWs is not well understood, especially in the presence of moisture. This study aims to comprehend the effects of timber-grain orientations and the moisture content (MC) on dominant GW modes, specifically, on the anti-symmetric A0 mode. The study is based on samples extracted from a Western White Pine utility pole, in three different orientations including longitudinal, tangential, and radial. The MC is introduced using a desiccator with a saturated salt solution. The studied MC conditions included dry, ambient MC, and high MCs varying from 10% to 27%. The GWs were excited using piezoelectric actuators and sensed with a Scanning Laser Doppler Vibrometer with Free-Free boundary condition. Phase velocities were obtained using the frequency-wavenumber plots of the measured line scans and were compared to the analytical curves for mode analysis. The results showed that the experimental A0-mode phase velocities match well with those generated analytically. The difference was negligible for longitudinally oriented specimens with a percentage difference ranging from 0.28% to 5.34%, as the MC varied. The phase velocity decreased by about 7% as the MC increased from 0% to 27%. A noticeable difference in the phase velocity was measured with different orientations; at dry conditions, the A0 propagated at 958 m/s in the longitudinal specimens, while it decreased in the tangential specimens to about 760 m/s, and had the lowest phase velocity in the radial direction reaching 433 m/s. The results show a high sensitivity of the propagating GWs to the direction of the fibers and the MC present in the timber.

Application of deep learning for structural health monitoring of a composite overwrapped pressure vessel undergoing cyclic loading

Structural Health Monitoring (SHM) using ultrasonic-guided waves (UGWs) enables continuous monitoring of components with complex geometries and provides extensive information about their structural integrity and their overall condition. Composite Overwrapped Pressure Vessels (COPVs) used for storing hydrogen gases at very high pressures are an example of a critical infrastructure that could benefit significantly from SHM. This can be used to increase the periodic inspection intervals, ensure safe operating conditions by early detection of anomalies, and ultimately estimate the remaining lifetime of COPVs. Therefore, in the Digital Quality Infrastructure Initiative (QI-Digital) in Germany, an SHM system is being developed for COPVs used in a hydrogen refueling station. In this study, the results of a lifetime fatigue test on a Type IV COPV subjected to many thousands of load cycles under different temperatures and pressures are presented to demonstrate the strengths and challenges associated with such an SHM system. During the cyclic testing up to the final material failure of the COPV, a sensor network of fifteen surface-mounted piezoelectric (PZT) wafers was used to collect the UGW data. However, the pressure variations, the aging process of the COPV, the environmental parameters, and possible damages simultaneously have an impact on the recorded signals. This issue and the lack of labeled data make signal processing and analysis even more demanding. Thus, in this study, semi-supervised, and unsupervised deep learning approaches are utilized to separate the influence of different variables on the UGW data with the final aim of detecting and localizing the damage before critical failure.

Wireless and battery-operatable IoT platform for cost-effective detection of fouling in industrial equipment

We present a novel internet of things (IoT) sensing platform that uses helical propagation paths of ultrasonic guided waves (UGWs) for structural health monitoring. This wireless sensor network comprises multiple identical sensor units that communicate with a host PC. The units have dedicated hardware to both generate and receive ultrasonic signals, as well as RF signals for use in triggering the sensors. The system was developed for monitoring and sensing pipelines and similar structures in real-time to facilitate interactive sensing. For accurate sensing with a limited number of arbitrarily scattered sensors, we obtain information from all sensor pairs and analyze helical propagation paths in addition to the commonly used shortest paths. UGWs can propagate long distances along the walls of pipelines, and their propagation velocity depends directly on the thickness of the waveguide, and is affected by energy leakage and mass loading. In this paper, we evaluated the network by utilizing it to detect fouling. The network could be adapted for further ultrasonic measurement tasks, e.g., measuring wall thicknesses or monitoring defects with pulse-echo methods.