European Workshop on Structural Health Monitoring

Full-field noncontact structural geometry morphology monitoring can be used to achieve a breakthrough in the fields of structural safety monitoring and digital twins owing to its advantages of economy, credibility, high frequency, and holography. Moreover, such type of monitoring can improve the precision and efficiency of the structural health monitoring technology and theory of large-scale structures. This study validated the performance of a proposed holographic visual sensor and algorithms in computer vision-based, full-field, noncontact displacement and vibration measurement. On the basis of the temporal and spatial characteristics of the measured series data, denoising, and the disturbance-rejection algorithm, the microscopy algorithm of subpixel motion and the extracting algorithm of motion information were respectively constructed for weak high-order displacement components and the holographic measurement of high-quality geometric morphology. Moreover, an intelligent perception method optimized for holographic-geometric and operational-modal shapes were used to extract morphological features from a series of holographic transient responses under excitation. Experimental results showed that the holographic visual sensor and the proposed algorithms can extract an accurate holographic displacement signal and factually and sensitively accomplish vibration measurement while accurately reflecting the actual change in structural properties under various damage/action conditions. The accuracy and efficiency of the system in the structural geometry monitoring for dense full-field displacement measurement and smooth operational modal shape photogrammetry were confirmed in the experiments. The proposed method could serve as a foundation for further research on digital twins for large-scale structures, structural condition assessment, and intelligent damage identification methods.

Interpretation of the data acquired from guided-wave-based measurements often utilizes machine learning. However, creating effective machine learning models generally requires a significant amount of data - which in the case of guided waves are costly and time-consuming to acquire. This limitation significantly reduces the application perspective of many advanced machine learning algorithms, most notably deep learning. The problem of data scarcity has been partially addressed in the field of computer vision via the usage of generative adversarial neural networks. These generate synthetic data samples, matching the real data distribution. Aside from images, generative adversarial networks have also been applied to synthesize audio data - with recent advances going as far as successfully synthesizing human speech. These developments suggest that they may be applicable for generating guided waves data - as fundamentally the problem is in many ways similar to that presented by audio waves. This work explores the capabilities of generative adversarial neural networks in the area of guided-wave signal synthesis. The used database was acquired in a series of pitch-catch experiments in which various sensor locations were utilized, and is significantly extended both in terms of sensor locations and data available from each sensor pair. Lastly, the resultant synthesized data is evaluated by qualitative signal comparison.

Structural health monitoring becomes popular and important in the field of structural engineering because this technology can elongate the structural life cycle as well as protect structures from natural hazards. In the past, structural health monitoring mostly relied on the contact sensors to acquire structural responses and then diagnosed structures from these measurements. Therefore, this study presents a deep learning-based method which can detect and segment the concrete cracks through the noncontact measurements, e.g., images. This method implements deep learning with computer vision to identify crack existence and to further perform segmentation. First, the training data are prepared by collecting images of concrete surfaces with/without cracks, and two state-of-the-art models such as DeepLabv3+ and Mask R-CNN are established along with the transfer learning methods and trained by real crack images. Then, the trained models can extract crack features and yield a mask (i.e. probability map). The cracks are identified and segmented in images from the predicted mask. Finally, the pixel-wise result is processed to determine the geometric properties of cracks such as lengths and widths. Three experiments are designed to examine these two models, and performance of these two models are evaluated with the mean intersection-over-union (mIoU) ratings. Moreover, a comprehensive comparison between Mask R-CNN and DeepLabv3+ is carried out. To sum up, cracks on the concrete surface can be successfully identified, and the near optimal selection of segmentation models under different scenarios is discussed and provided in this study.

This study proposes a deep-learning-based condition assessment approach by reconstructing marginal probability density distributions of girder vertical deflection (GVD) and cable tension (CT) using unsupervised image transformation (UNIT) model for a real cable-stayed bridge. 27 and 139 sensors of GVD and CT are distributed along the full bridge, and both the sampling frequencies are 2 Hz. First, vehicle-induced components of GVD and CT are extracted by data pre-processing while the temperature-induced components are removed, a time window of 3 h with a sliding length of 10 min is used to obtain original GVD and CT segments, and the corresponding margin probability density distributions (PDFs) are calculated by kernel density estimation. The training and test sets consist of 2986 and 6236 PDFs, respectively. The proposed UNIT model consists of a series of variational autoencoders (VAEs) and generative adversarial networks (GANs). Both the PDFs of GVD and CT are used as inputs. Then, the proposed UNIT model is updated by solving the mini-max problem of training GANs, in which VAEs act as generative models. Finally, the Wasserstein distance between the predicted and ground-truth PDFs of CT acts as the indicator of bridge condition change. Results show that specific modes of PDF variations induced by SHM system upgrade, cable damage, data anomaly, and traffic jam can be recognized to assess bridge condition.

The development of urbanzation areas requires more land and space in the already congested space. Increasingly new structures are having to be constructed in proximity to existing underground infrastructure. In the UK, this infrastructure is of varying ages and the new construction poses a potential threat to the serviceability or integrity of existing structures in respect of their original SLS and ULS design capacities.One of the concerns is construction vibration. Activities such as piling, demolition and compaction will generate vibrations at various frequencies. Due to the complexity of vibration propagation related to the structure damage, there is a lack of reliable data on the threshold of vibration-induced damage in structures both in countries where national standards already exist and in the UK. This paper will summarize the currently available UK guidance, BS 5228-2, BS 7385-2, and CIRIA TN142, build up a systematic assessment approach, and present real monitoring data collected from construction sites. These monitoring data are normally enormous and recorded in real time, hence opened the opportunities for artificial intelligence data transmission and alert notification.The assessment approach suggested to adopt the 3 Tiers system similar to the better developed building response to ground movements assessment proposed by Burland (1995) and Mair (1996). The first tier will screen out any structures monitored to have less than a threshold of vibration because damage to these structures is highly unlikely. The second tier will look at individual sources of vibration frequency, if applicable, and their respective limits will be assigned according to available case histories. The third tier will apply if monitored vibrations at a particular structure exceeds the limits set out in the second tier. A detailed inspection and specific structure assessment will be required to ensure the integrity of such a structure. The artificial intelligence driven monitoring can apply this criterion accordingly.

The aim of this paper is to develop a grey-box modelling method that incorporates both physical insight and learning from data. In this paper, a simple and intuitive approach is adopted, which is to represent physical laws through the prior mean function in a Gaussian process. This approach naturally adheres with the Bayesian viewpoint. The benefits of the approach, particularly in extrapolation, are shown in an example case study, where a model to predict deck displacements of a cable-stayed bridge is developed.

The work is devoted to the detection of damage in the fuselage lap joints through strain monitoring. The need to analyze the deformations at a large number of points is one of the most important problems when using the strain gauge data to monitor the state. In this work, the method was proposed to determine the optimal monitoring parameters based on tensometry data. The work is based on the analysis of the interrelation between the monitoring system readings and the fatigue damage in the joint elements.The dependence under study is presented as a neural network (NN), which is obtained based on processing the results of full-scale tests, as well as by analyzing the local deformations in the joint finite element (FE) model, taking into account the various options for the cracks location. The study of cracks and determination of their safe size was carried out based on the results of fractography of the longitudinal joint samples. To determine the optimal parameters for monitoring the fuselage joint elements, the problem is solved to find the conditional extremum of function, which is presented in the form of neural network approximation.

Identification of elastic properties is relevant for both non-destructive material characterizations as well as for in-situ condition monitoring to predict any possible material degradation. In this paper, we have proposed deep learning frameworks to solve the inverse problem of material property identification using ultrasonic guided waves. The propagation of guided waves in a composite laminate is modelled using a reduced order Spectral Finite Element Method (SFEM). We have used two fundamental modes of a guided wave i.e. the anti-symmetric (A0) and the symmetric modes (S0) as inputs for the proposed deep learning models and elastic properties of a unidirectional composite laminate as outputs. The deep regression-based networks like, 1D-Convnets and LSTMs are used to map input space to target space. The performance of the algorithms is evaluated based on the mean squared loss value, coefficient of determination, and mean absolute error. It is seen that the networks can learn the inverse mapping and generalize well to unseen examples even in the presence of noise at various levels. This novel methodology can eliminate disadvantages associated with existing global optimization techniques in terms of accuracy, robustness, scale, and computational time.

For optimal performance, the machine learning algorithms requires enough number of training data. The major issue with these algorithms in damage diagnosis is unavailability of data contains range distinct damaged scenario recorded from a real structure. The generation of such data experimentally is a challenging task. To address this issue, a methodology is proposed by which the diagnosis algorithm is trained with numerically simulated database and tested on real experimental data. The proposed methodology consists a unique feature extraction procedure that extract the features from simulated data which are close to experimental data. The methodology is employed for quantification and localization of debonding in a metallic stiffened panel using the vibration-based approach. The natural frequency of the undamaged numerical simulated model is experimentally validated; the validated model used to create numerically simulated damage database. The feature is extracted from first mode displacement raw data. The Artificial Neural Networks (ANNs) have been chosen as diagnosis algorithm. The algorithm structure has been optimized on numerically simulated database. Finally, the feasibility of the algorithm is tested by processing real experimental data trough optimized algorithm recorded by Laser Doppler Vibrometer (LDV). The proposed methodology is produced a good prediction accuracy in quantification and localization of debonding on experimental data.

Acoustic Emission (AE) is a non-destructive technique suitable for monitoring structural damages in complex systems. Interpretation of AE data is generally made by an unsupervised learning approach called clustering. A clustering method divides a dataset into different groups called clusters, which are expected to have a physical interpretation in terms of damages. The set of groups, called partition, is then evaluated through an external criterion. Since the number of clusters is a priori unknown, the standard approach for AE clustering consists in running a clustering method for different number of clusters and then to select the partition with the best value of the external criterion. The external criterion can vary across publications and aims at evidencing “natural” clusters. According to the criterion, the selected partition and its interpretation can be subjected to variability. In this publication, we explore a clustering method in which the parameters of clusters are optimised jointly with the number of clusters through an objective function and an optimisation procedure designed for that purpose. To our knowledge, it is the first method of this class in AE literature. Experimental results concern real data from tightening test.

The paper presents recent results of diagnostic and monitoring activities carried out on the Consoli Palace of Gubbio (Italy) through satellite radar interferometry and in-situ continuous measurements. The work aims to investigate the effectiveness of remote sensing technologies in the understanding of complex phenomena, such as seismic vulnerability aspects concerning crack amplitudes evolution over time and soil-structure interaction, for the conservation of Cultural Heritage (CH). These activities have been carried out within the European HERACLES project, funded in the framework of Horizon 2020 and aimed at proposing novel diagnostic monitoring solutions for enhancing heritage resilience against various types of hazards. As far as it concerns the radar interferometry analysis, two datasets from the COSMO-SkyMed satellite constellation were used, processed with the Persistent Scatterer Pair (PSP) and SAR Tomography (TomoSAR) techniques, while the on-site structural monitoring system consists of two LVDTs installed on the Consoli Palace, each of them equipped with environmental sensors, in particular with thermocouples. The comparison between the outputs of satellite radar interferometry and in-situ structural monitoring data of the two test-beds, made between July 2017 and the end of 2018, allows assessing the potential of remote sensing techniques in supporting decision-making and proactive interventions for CH conservation.

In the last decades, the Synthetic Aperture Radar (SAR) technology has undergone a quality increase in signal acquisition and therefore a greater quality and precision of the measurements. This allowed to develop a technique capable of capturing the displacements over time, called Differential Interferometric SAR (D-InSAR). At first, the Satellite Interferometry (InSAR) technique was used to determine the displacements that occurred as a result of earthquakes, land-slides and volcanic eruptions. Now, thanks to the sensitivity achieved, several studies are currently ongoing to investigate the possibility of using these techniques to monitor the movements and deformations of buildings and infrastructures. In this paper, an attempt is made to evaluate the correlation between DInSAR data and the damage states of buildings based on differential settlement scenarios. For this purpose, nonlinear numerical models of RC buildings have been developed and analyzed.

In the present work, an Advanced Differential Interferometric Synthetic Aperture Radar (A-DInSAR) technique has been used to detect the displacements of the upstream tailing dam of Vale’s Córrego do Feijão. The study area is located to 9 km east of Brumadinho, Minas Gerais, Brazil, where on 25 January 2019 the Dam was affected by a catastrophic failure causing a huge flood that destroyed the mining officers, houses, roads and 257 people died as a result of the collapse. The dam was built in 1976 by Ferteco Mineração using the upstream heightening method, the latter consists to build the dam body using the deposited tailings. The Vale’s dam height was 86 m and its crest length was 720 m. The tailings occupied an area of 249.5 thousand square meters and the disposed volume was 11.7 million cubic meters. The study has been carried out using the 64 descending satellite images acquired as Single Look Complex (SLC) from Europe Space Agency of the Copernicus Programme. The data have been processed by means of Coherent Pixels Technique (CPT) algorithm. The result consisted of a map of displacements on the body dam from the 2016 to 2019, in the period before the collapse. Furthermore, the study of the time series of displacements allowed to assess the evolution of dam structure deformations. The study confirms that the A-DinSAR technique could be very useful tool for monitoring tailings dam and to reduce the risks.

The prevention and mitigation of slope instability requires effective technologies to reduce the vulnerability of existing structures. Landslides are global phenomena, caused by natural geological phenomena or induced by anthropogenic sources. Slow and intermittent landslides result in a significant number of losses, as well as physical damage and extensive economic losses to private and public property. The physical vulnerability of buildings to landslides is a term used to describe their potential for physical loss when they are affected by movements induced by unstable terrain. Therefore, monitoring plays a key role in the management of natural hazards and assumes a fundamental task to provide cost-effective solutions to mitigate or minimize physical and economic losses. Remote sensing techniques proved to be powerful investigative tools due to their high spatial and multi-temporal coverage, rapid data acquisition and overall reasonable costs. The main aim of the present issue is to provide a general methodology that can be used to predict the spatial and temporal evolution of a slow landslide. In particular, this work analyses a slow-moving phenomenon which affects the University of Azuay (Cuenca - Ecuador). Multi-level analysis integrated with innovative monitoring techniques and geophysical analysis methods lay the foundations for greater accuracy for the prevention and prediction of such phenomen.

Data-driven methods are widely used in structural health monitoring (SHM) systems, and most of them focus on feature parameters extracted from damaged structure. However, structure is usually in healthy situation, which produces much more healthy data compared with damaged data. The classification of severity of damage using both healthy and damaged data is an imbalanced classification problem. This paper presents a damage classification method using Lamb wave and Synthetic Minority Over-sampling Technique with Iterative-Partitioning Filter (SMOTE-IPF) algorithm. A fatigue test with Lamb wave detection is conducted and three damage sensitive features, namely, normalized amplitude, correlation coefficient and normalized energy are extracted from signals as dataset. Generation of minority examples and filter of noisy examples are implemented with SMOTE-IPF method. Cross validations are performed on the proposed model using feature parameters and the length of fatigue cracks. The metric parameter for classifier performance is calculated to verify the performance of the proposed method for crack size identification.

Future space structures, while obeying the low-mass-to-orbit requirement, are expected to drastically increase their size. As redundancy in structural elements is typically not allowed, the risk of failures increases, and information about damaged parts could be useful to avoid excess loads. Furthermore, as distributed active control is likely to be included, new schemes could be envisaged to bear with detected failure(s). Identification of weakened elements in these large structures is indeed of paramount importance. Bayesian estimators, intrinsically capable to work with noisy data provided by sensors, can be a suitable tool for monitoring the health of structure. The paper investigates the performance of these estimators for failure detection and identification problems referring to a rich, realistic model of a future large radar satellite. Multiple models built on different possible behaviors are considered together to timely set the alarm when the likelihood threshold for a possible failure is passed. The capability to identify differently located occurrences is analyzed, discussing the confidence in the solution. Aside from the well-known literature on Bayesian estimators, focus is on the hints which could be gained from realistic simulations in view of the possible operational applications to space structures.

Guided electromagnetic waves (GEW) in the GHz frequency range have been introduced recently as a novel approach for monitoring surface damage of metallic structures. Therefore, a rectangular waveguide or a dielectric waveguide patch can be attached to the structures’ surface. Together with the host structure these two components form a conventional electromagnetic waveguide by which surface changes coming from defects can be detected and monitored sensitively.While the focus of previous research was on damage detection and damage localization in 1-dimensional microwave waveguides, we present here a damage localization technique in the frequency band from 30–40 GHz for multiple transmitters and multiple receivers (MIMO approach) that are permanently attached to the structure. For this goal, all transmitter-receiver pairs have been simulated in CST Microwave Studio for the baseline state and the damaged state of the structure. The numerical simulation contains flat bottom holes at different spatial locations of the plate. The receiver signals have been processed by means of well-known tomography and digital beamforming algorithms to determine the damage location accurately. Finally, a comparison of the image quality has been implemented.

In the railway field, the safety of passengers and the service life of train components are a crucial issue. For this reason, continuous periodic inspections by non-destructive techniques are required. Among these, ultrasonic tests are widely used in this field, even though the conventional ultrasound techniques have the disadvantage of requiring the disassembly of the wheels and of putting the train out of service. This procedure is expensive and time-consuming and can be neglected if non-contact ultrasonic techniques are used.In this work, the authors present an experimental research on some defects, artificially obtained on a railway wheel supplied by Trenitalia Spa, by adopting three different experimental setups and comparing the results. The proposed method uses laser equipment in generation and reception of the ultrasonic waves. The receiving unit combines a continuous wave laser and an interferometer in order to detect the surface displacements. In addition, when used in service conditions, the receiving by the interferometer is facilitated, because the tread of the wheel is very reflective (thanks to wheel – rail contact).The research aims at optimizing the experimental setup through a number of measurements carried out by varying some experimental parameters to evaluate the robustness and reliability of the technique.

In this paper, we present a simple predictive model to estimate the natural frequencies of steel cantilever beams with branched cracks. The model considers the cracked beam as a beam with constant cross-section, on which acts an equivalent bending moment. This bending moment takes into account the modification of the rigidity on the beam segment on which the crack extends in the longitudinal direction and the additional rotation that occurs in the slices located at the ends of the damaged segment. Along with the theoretical background and the deduced mathematical model, we present several cases of deterioration defined by the depth of the transverse crack branch respectively the position and the extension of the longitudinal branch. These examples prove the accuracy of the predictions regarding the change of the natural frequencies due to the damage obtained with the proposed model.

Since the dawn of humanity, nature has been a source of inspiration for developing engineering systems, referred to as “nature-inspired systems”. With respect to smart structures instrumented with smart structural health monitoring (SHM) systems, nature-inspired systems may provide promising advancements, for example, by executing self-healing or self-diagnosing processes. However, for developing optimum strategies towards deploying nature-inspired systems to smart structures, the plenitude of nature-inspired systems in SHM need to be classified. This paper aims at reviewing the potential of nature-inspired systems to advance the performance of smart structures. Upon a brief introduction to smart structures and nature-inspired systems, a state-of-the-art review of nature-inspired systems that exhibit potential to advance smart structures is presented, providing decision support on how to advantageously apply the benefits of nature-inspired systems to smart structures.

Cultural heritage protection and restauration are fundamental matters. Intervention design requires preliminary modelling and analysis to carefully simulate the structural behaviour of existing buildings. The identification of constructive schemes is based on direct surveys, whereas direct testing are required to reveal mechanical and physical properties of materials and their degradation status. Clearly, higher knowledge levels correspond to minor penalties in terms of material performances. For metal structures, regulations provide the employment of destructive investigations only. Furthermore, the sampling of specimens often collides with the safety requirements of artifacts. Therefore, there is a strong need for non-destructive investigations, such as the Leeb method, for a reliable in-situ characterization of carpentry steels. A fundamental step towards reaching this aim is represented by the identification of a theoretical relationship between Leeb hardness values, measured in-situ, and experimental tensile strengths. In order to identify a generally valid correlation, data of the past four years were collected from the database of the Tecnolab s.r.l. company. The experimental setup was based on in-situ Leeb analysis followed by samples collection for consequent tensile tests performed in the laboratory. The experimental data, compared to the trend provided by internationally valid guidelines, provide resistances that the regulations tend to overestimate. Therefore, designing an intervention using these resistances would not be on the safe side. Further analyses should be performed to investigate determinants related to in-situ conditions altering the steel resistance with the aim of identifying potential corrective factors.

Engineers typically approach monitoring design differently from structural design, even though they are logically equivalent processes. In designing a structure, engineers follow a well-established rational procedure, whereby the performance of the design concept is predicted through structural analysis and quantitatively compared with the design target. In contrast, in designing a monitoring system, their approach is often heuristic with performance evaluation based on experience or common sense, rather than on quantitative analysis. The reason is that there is not a broadly accepted method for the design of monitoring systems. We propose a quantitative method for the calculation of the expected performance of monitoring solutions and for checking their effectiveness, called performance-based monitoring system design. It is the counterpart of the semi-probabilistic structural design and is based on the calculation of monitoring capacity and monitoring demand – the counterparts of structural capacity and structural demand. Capacity and demand are defined as the uncertainty of the structural state that will be estimated through the monitoring system, with a difference: the capacity is the uncertainty that affect the actual estimation; the demand is the maximum uncertainty acceptable to make the system effective. After formalizing the method, we apply it to a real-life monitoring problem, which shows with real numbers the calculation of the monitoring capacity. Performance-based monitoring design can reduce the monitoring costs and improve the monitoring effectiveness, allowing infrastructure operators to identify the condition state of a structure with the desired accuracy.

The assessment of Cultural Heritage buildings is a challenging multi-disciplinary activity, involving different tasks. The paper exemplifies the application of the methodology involving historic and architectural research complemented by a dynamic survey in the structural modelling of the Galleria degli Antichi, a monumental building built in the 16th century in the historic town of Sabbioneta (Italy), which is included in the UNESCO World Heritage list since 2008.After a concise historic background on the investigated building, the paper focuses on the results obtained from operational modal testing and the subsequent finite element (FE) model correlation analysis.The good knowledge of the structural geometry and the large number of identified vibration modes, combined with a classic system identification technique, allowed to establish a linear elastic FE model, accurately fitting the modal parameters of the monument in its present condition.

In the online approach for Structural Health Monitoring (SHM) particular relevance can be assumed in the Artificial Intelligence algorithms. In fact, the real time structure condition evaluation must be performed by the fast analysis of the data provided by several sensors and devices. Among the SHM techniques the most attractive are based on the analysis of the Acoustic emission. In particular, in the paper are considered the acoustic emission naturally generated by the building material under stress. This signal is used in the paper to evaluate the input features for the Machine learning. The features of the acoustic emission signal used in the training of the machine learning techniques are based on the Gutenberg–Richter law, which expresses the relationship between magnitude and total number of earthquake events in a defined region and time interval.In the design of the automatic classifier of structural critical event the selection of the machine learning technique is performed by comparing with tests the performance of the: (i) Support Vector Machine (SVM), (ii) k Neighbor Classifier and (iii) Adaptive Boosting. The experimental tests performed by compression tests carried out on cubic concrete specimens permit to achieve the best performance as concerning the overall classification accuracy.

Asset inspection of large structures such as storage tanks in the oil, gas and petrochemical industry is challenging, either requiring labour-intensive manual measurements or using robotic deployment to make the measurements. Current robotic systems employ point-by-point scanning, which is time-consuming. Using guided waves for such inspections is attractive as they provide a mechanism for monitoring the inaccessible areas and simultaneously providing structural location data to speed up the inspection process. In this research, shear horizontal (SH) guided waves generated by electromagnetic acoustic transducers (EMATs) are used to screen a large area using a crawler. EMATs with 22 mm wavelength are used to generate the first two SH modes: a non-dispersive SH0 and highly dispersive SH1 on a 10 mm thick steel sample. Previously, we have demonstrated the feasibility of guided wave-based occupancy grid mapping (GW-OGM) for mapping a structure’s edges. In this work, the GW-OGM technique is generalised to identify and estimate the location of a flat bottom hole in a pitch-catch mode. The simulation and empirical data demonstrate that the location of damage can be identified as the robot navigates on the component, with full coverage. Moreover, the simulated data are in good agreement with the experimental results on the generation of SH wave modes.

A radar equipment was used to measure the deflection response of bridge stay-cables induced by ambient and traffic excitation. After a concise description of the radar equipment and a summary of advantages and potential issues of the microwave technology, the paper focuses on the experimental tests performed on all stay-cables of the curved cable-stayed bridge erected in the commercial harbor of Porto Marghera, Venice, Italy. The bridge consists of an inclined concrete tower, single-plane cables and a composite deck; the curved deck has a centerline length of 231 m, with two different side spans and 9 cables supporting each side span.Three series of ambient vibration tests were performed (on July 2010, April 2011 and October 2019) on the two arrays of cables of the bridge by using conventional accelerometers and microwave interferometer. The availability of simultaneously collected radar and accelerometer data (which are usually regarded as reference data in dynamic tests) allowed to investigate the accuracy of the radar technique (in terms of natural frequencies and tensile force estimated from natural frequencies) and the errors/uncertainties in radar results. Furthermore, the tests allowed to verify the repeatability of radar survey, with SHM purposes.

Structural health monitoring of electric furnaces is a feasible way to achieve maintenance goals without momentary stoppages and unnecessary repairs. Monitoring of refractory lining erosion via brick thickness is implemented within a furnace. However, other techniques are necessary in case of damage of the original system. NDT techniques, especially thermography, are tested in this work. Several thermograms were taken in different panels around the furnace. A basic heat transfer model is used to calculate brick thickness using composed images of each of the considered panels. Results show that this technique can predict brick thickness, considering possible error measurements in one of the auxiliary variables.

Thermography is a well-known technique in industrial applications to check for refractory lining deterioration. In this work, experimental temperatures are obtained for different panels around an electric furnace for ferronickel smelting, and heat from each panel or zone is calculated for both normal operating conditions and during a shutdown and starting. Results showed that thermography can spot differences in surface temperature and heat transferred between panels, indicating that refractory is eroded in the ones with bigger values. Thermography can also follow quite well the transient behavior of the furnace.

The anisotropy and heterogeneity of composites influence unavoidably the mechanical response of the material to external excitation and the failure mechanisms. As an effect, the mechanical behaviour assessment of composites by means of experimental techniques requires to pay attention to the influence of the specific layup of laminae, viscous properties of the matrix, pattern described by the yarns or fibers. It follows that specific quantitative and qualitative analysis are required for the data processing.The study of thermal signal can be a successful strategy to assess and to understand the damage processes. In effect, if compared to other experimental techniques, it allows a localised analysis of the material degradation in terms of stiffness reduction or damage progression (transverse cracks or delamination).The present research is aimed at providing innovative methods and algorithms for processing the thermal signal of a composite obtained by Automated Fiber Placement process, in order to determine when and where damage occurred during static and cyclic loading in terms of transverse crack number.

Adhesive joints represent a valid alternative to classical welded and bolted methods, above all for composite materials in which the weight of structures plays an important role for designers. Defects such as debonding of adherents can compromise the correct static and dynamic behaviour of adhesive structures. In this regard, NDT techniques can be used for detecting and monitoring defects and damaged areas during the operative loading conditions. In this work, the capability of the Thermoelastic Stress Analysis (TSA) technique of monitoring debonding in adhesive T-joints has been investigated. Moreover, a quantitative comparison with the well-established Lock-in thermography and Ultrasounds techniques has been provided.

The paper presents the concept and application of the smart pedestrian footbridge, equipped with DFOS strain sensors called EpsilonRebars. These sensors, in the form of composite rods being simultaneously the structural reinforcement for the concrete deck, were placed along the entire span of nearly 80 m. Thanks to the application of distributed optical fibre sensing technique DFOS, it is possible to perform measurements of strains, displacements (deflections) and temperature changes in a geometrically continuous manner along the entire length of the footbridge. The sensors integrated with the deck were used to measure selected physical quantities during the hydration of early-age concrete (thermal-shrinkage strains) as well as during the load tests. Sensor readings can be performed at any time of the structure operations in order to assess its technical condition (e.g. crack appearing) and to analyze the impact of environmental conditions and other factors, e.g. rheological phenomena.

The inverse Finite Element Method (iFEM) is an algorithm able to compute the deformed shape of a structure starting from its geometry, the boundary conditions, and the strain data measured on discrete positions of the structure, without prior knowledge on material properties and load condition. A critical point of this technique is the sensor pattern optimization for a correct strain field reconstruction. Sensors cannot be applied on the whole structure due to the presence of the physical boundary conditions and logistic constraints, potentially leading to a wrong reconstruction of the deformed shape by the iFEM. Thus, the Smoothing Element Analysis (SEA) is used in this work for prior extrapolation of the strain field in locations where sensors are not available. A sensitivity analysis highlights the influence of the mesh size and different SEA hyper-parameters over the extrapolated strains, which are then provided as input to the iFEM for shape sensing. The method has been verified with experimental tests on a CFRP specimen sensorised with fiber optic, subjected to a compressive load, proving the iFEM can compute the deformed shape of the structure also in presence of a buckling condition. The iFEM results are validated with experimental displacement measures from lasers.

The Environmental and Operational Conditions (EOC) measurement and compensation in Structural Health Monitoring (SHM) systems for event and damage diagnosis during the aircraft operation is a key aspect to ensure their reliability. In addition, those EOC measurements can be used to support the existing usage and operational loads monitoring systems of Airbus Defence and Space (ADS), enhancing its capabilities.Previous works have been addressed in flat reinforced composite panels in the framework of ADS SHM diagnostic system for events and damages development. This paper faces the complexity of implementing this system in a large composite fuselage structure installing a hybrid SHM system. It is based on the combination of dynamic distributed fibre optic sensors (DFOS) and piezoelectric sensors measurements with the target of diagnose events and damages considering the EOC.The results show the advantage of using hybrid combinations of sensors to ensure the reliability of SHM diagnostic systems in real operation

The use of optical fiber sensors (OFS) has spread in the Structural Health Monitoring (SHM) community for their ability to detect many different physical quantities, robustness against electromagnetic disturbances, light weight and embedding possibilities. The last point has been widely investigated for different types of materials, but only recently researchers considered the possibility to embed optical fibers in 3D printed structures. Additive Manufacturing (AM) offers new opportunities in terms of design, for the manufacturing of structures with complex geometries in a relatively low amount of time. However, new challenges must be considered, including innovative embedding solutions for different types of sensors. As a first step, this work discusses current embedding strategies for optical fiber sensors in structures produced with the Fused Deposition Modeling (FDM) technique. A novel methodology to embed OFS is introduced and then tested through the production of specimens at three different filling densities and six different loads. The experimental results, where both distributed OFS and strain gauges were used, were also compared with the data obtained from a numerical model developed in Abaqus/CAE in which the filling pattern of the specimens was accurately reproduced. Finally, the results were critically discussed, highlighting both agreements and discrepancies with respect to the expected data.

Continuous and proper operation of valves is essential for reliable operation and safety of power and petrochemical plants, especially nuclear power plants. Two of very important operating parameters of a motor operated valve (MOV) need to be measured are the torque and position of a valve stem produced by an actuator. A method of employing a fiber Bragg grating (FBG) sensor embedded into the surface of a worm shaft by epoxying to measure the axial force induced in the worm is proposed for measuring the torque and position of an MOV actuator which is indicative of the torque and position of a valve stem, where the axial force is proportional to the torque and periodically changes with the worm rotating. The strain characteristics of the FBG sensor are evaluated by uniaxial tensile testing to assure adequate strain transfer to the grating. The response of the sensor to the axial loads acting on the worm is linear with high repeatability and strain transfer efficiency. A test system capable of inducing a variety of adjustable torsional loads on an actuator is provided for determining its torque and position. The response of the sensor to the actuator output torque is also linear, while the periodical changes in the Bragg wavelength of the sensor during actuator operation make it possible to obtain the position of a valve stem simultaneously. The results demonstrate that this method provides a great potential for simultaneous measurement of the torque and position of valves in power and process industries.

In this paper, the identification of impact area that works only with the readout responses opportunely post processed, is proposed. The readouts signal is representative of a distributed high resolution strain map under backscattering random noise and static excitation.The algorithm is based on first order derivative operator already developed by the author. Up to four features are extracted from time and spatial domain. Then the cross-correlation function is applied to improve signal to noise ratio. A basic triangle cluster applying the Delaunay algorithm for area reconstruction is implemented. The methodology is applied to estimate the surface extension of an impact damage induced by low/medium energy impacts on 24 and 36 ply CFRP panels with non-unidirectional layups. The results show a good coherence with respect to the ultrasonic NDI controls.

Monitoring on engineered barrier of radioactive waste repository is an issue of safe operation and staged closure. Optical fiber sensor, which is based on back-scattering, is expected to reduce a negative impact on the barrier performance because of multi-channel sensor. In Japanese concept of intermediate depth disposal, a cement wall is expected to work as a low diffusion layer of the engineered barrier system, and thus crack on the wall is critical for the ensure of its function. To monitor the occurrence and development of crack on the surface of the concrete structure, optical fiber is mounted on the surface of full-scale mockup and has been evaluated through a cracking test in underground research laboratory. The authors apply both Brillouin-scattering- and Rayleigh-scattering-based optical fiber sensor to measure strain distribution when crack generates, and then the pros and cons of each sensor are experimentally clarified in terms of crack monitoring. The results indicated that Rayleigh-scattering-based measurement can predict the stress concentration before cracking, and Brillouin-scattering-based measurement can trace the crack opening after cracking. Based on measurement in situ with the use of optical fiber sensor, it can be said that crack monitoring on cementitious diffusion layer is feasible, and then integrating different scattering sensor is proposed for wide coverage of measurement.

Damage detection on an operating wind turbine blade is, for three progressive scenarios of 15, 30, and 45 cm long trailing edge damage, considered via a single vibration acceleration sensor. The signals are from 3–month–long measurement campaign, involving varying environmental and operating conditions and uncertainty with significant effects on the dynamics, which almost completely ‘mask’ the effects of damage. The study employs two vibration–based detection methods: A conventional Unsupervised AutoRegressive model based (U–AR) method and a robust Unsupervised Principal Component Analysis Multiple Model AutoRegressive model based (U–PCA–MM–AR) method. The results of the study confirm the inadequacy of the U–AR method, yet the high performance achieved by the U–PCA–MM–AR method, which approximates well that of an 8–accelerometer–based counterpart. The study thus suggests that high detection performance is, especially for the two ‘larger’ damages, achievable via a significantly reduced set of sensors.

In data-driven SHM, the signals recorded from systems in operation can be noisy and incomplete. Data corresponding to each of the operational, environmental, and damage states are rarely available a priori; furthermore, labelling to describe what each the measured signals represent is often unavailable. In consequence, the algorithms used to implement SHM should be robust and adaptive, while accommodating for missing information in the training-data – such that new information can be included if it becomes available. By collecting three novel techniques for statistical learning (originally proposed in previous work) – including semi-supervised, active, and transfer learning – it is argued that probabilistic algorithms offer a natural solution to model the signals recorded from systems in practice.

Vibration-based Structural Health Monitoring (SHM) methods often rely upon vibration responses measured with a pervasive network of sensors. In some cases, it does not look possible for technical and economic reasons to equip civil structures with a distributed sensing system. Hence, the amount of information to handle for damage detection may be seriously affected by environmental and/or operational variability, leading to false detection results. To address this challenge, we present a parametric spectral method based on AutoRegressive (AR) modeling to set the damage-sensitive structural features. The spectra of the AR models associated with the normal and damaged conditions are collected into two matrices, to provide individual multivariate feature datasets in the frequency domain. By vectorising the matrices, two series of feature samples relevant to the normal and damaged conditions are obtained. To detect damage, the Log-spectral distance method is adopted to measure the similarity between the two aforementioned feature vectors. The effectiveness and accuracy of the proposed approach are assessed through limited vibration data relevant to the IASC-ASCE benchmark problem. Results show that the AR spectrum stands as a reliable and sensitive feature for partially observed structures, hence in the case of limited sensor locations; additionally, the presented distance methodology succeeds in detecting early damage.

Structural Health Monitoring (SHM) has brought various benefits into the industry, such as in economic, life-safety and lightweight design aspects. The recent development of SHM techniques leads to the optimization of the sensor arrangement, enhancement in the damage identification accuracy and increases in computational efficiency. Vibration-based SHM has shown to be practical and requires a relatively small number of sensors. In this approach, the dynamic responses at vibration nodes (node displacement, or NODIS) are adopted for damage identification in plate structures. The method utilizes the framework of Bayesian inference to overcome the drawbacks of traditional deterministic approaches so that the presence of various uncertainties and errors can be taken into consideration. Furthermore, the Markov chain Monte Carlo sampling technique with adaptive Metropolis algorithm is integrated into the framework, achieving a considerable reduction of computational costs. In this paper, the principles of the NODIS method for plate structure is elaborated. Then, a thorough explanation of the MCMC-based Bayesian framework with NODIS and its theoretical background are presented. At last, the performance of the proposed method is evaluated with numerical results.

Steel storage pallet racks are framed steel structures commonly used in the logistic field. According to the European practice, they are built with cold-formed steel profiles. Vertical and horizontal elements are connected with mechanical joints and special elements are used for the base connections. The design of these structures is usually performed by adopting the ‘design by testing’ approach. This procedure asks for the experimental characterization of the main racks components and sub-assemblies, which allows identifying the parameters needed for the safety checks and the development of reliable FE models. Recent seismic events clearly showed the need for improvements in the knowledge of the seismic response not only of the components but also of the whole structure. As a contribution to this topic, an experimental study of the seismic response of full-scale rack frames is currently in progress. At this aim, a testing set-up for full-scale structures, with a maximum height of 22 m, was designed and realized. In this paper, the main features of the experimental set-up and the results of two push-over tests on a commercial two-bay four-level pallet rack are described and discussed. Finally, the results of FE analyses are presented.

Active-sensing acousto-ultrasound structural health monitoring (SHM) constitutes an important family of methods for both metallic and composite structures. However, the presence of varying operational and environmental conditions in the real world can significantly affect their accuracy and robustness in the face of uncertainty. In this context, statistical learning methods that can be based on Gaussian Process regression models (GPRMs) and statistical time-series models can be incorporated in the damage diagnostic process to account for and properly represent such uncertainties. Towards this end, the main objective of this paper is the postulation and experimental assessment of two statistical learning approaches, based on GPRMs and time-varying time series models, for active sensing SHM under varying structural and loading states under uncertainty. The proposed methods involve GPRM representation of the non-linear mapping between the actual states with (i) traditional damage indices (DIs) and (ii) parameters of time-dependent autoregressive (TAR) models. The experimental validation and comparative assessment is based on a series of experiments on an aluminum coupon outfitted with a network of piezoelectric actuators/sensors subjected to different static loads under increasing damage size.

Structures and machines maintenance is a hot topic, as their failure can be both expensive and dangerous. Condition-based maintenance regimes are ever more desired so that cost-effective, reliable, and damage-responsive diagnostics techniques are needed. Among the others, Vibration Monitoring using accelerometers is a very little invasive technique that can in principle detect also small, incipient damages. Focusing on transient faults, one reliable processing to highlight their presence is the Envelope analysis of the vibration signal filtered in a band of interest. The challenge of selecting an appropriate band for the demodulation is an optimization problem requiring two ingredients: a utility function to evaluate the performance in a particular band, and a scheme to move within the search space of all the possible center frequencies and band sizes (the dyad {f, Δf}) toward the optimal. These problems were effectively tackled by the Kurtogram, a brute-force computation of the kurtosis of the envelope of the filtered signal (the utility function) of every possible {f, Δf} combination. The complete exploration of the whole plane (f, Δf) is a heavy task which compromises the computational efficiency of the algorithm so that the analysis on a discrete (f, Δf) paving was implemented (Fast Kurtogram). To overcome the lack of robustness to non-Gaussian noise, different utility functions were proposed. One is the kurtosis of the unbiased autocorrelation of the squared envelope of the filtered signal found in the Autogram. To spread this improved algorithm in on-line industrial applications, a fast implementation of the Autogram is proposed in this paper.

The feasibility of on–board hollow worn wheel detection in railway vehicles moving under different speeds based on bogie random vibration signals, is for first time investigated in this study. Towards this end, two unsupervised Statistical Time Series (STS) methods which are founded on Multiple Models (MM) for the representation of the vehicle partial dynamics are employed: The Unsupervised–MM–Power Spectral Density (U–MM–PSD) and the Unsupervised–MM–AutoRegressive (U–MM–AR). The methods’ assessment is achieved based on thousands of test cases from field tests with an Athens Metro railway vehicle possessing early stage hollow or reprofiled wheels under two nominal speeds (70 or 80 km/h). The results indicate both methods’ excellent performance based on vibration measurements on the lateral direction, as well as the increased sensitivity of the U–MM–AR method to the detection of hollow worn wheels when vertical vibration signals are used.

Digital Image Correlation (DIC) techniques have significant advantages in structural health monitoring compared to local measurements, being a non-contact method and providing distribution of engineering response indicators such as strains and displacements. In this paper, DIC is applied to investigate the bond behavior between Externally Bonded (EB) Fiber-Reinforced Cementitious Matrix (FRCM) fabric and concrete substrate. This is a popular strengthening technique for retrofitting existing concrete structures. Considering the different failure modes observed at the interface (e.g. cohesive debonding of concrete substrate, detachment at matrix-to-substrate interface, fiber sliding within the matrix, fiber rupture, etc.), a critical issue is to investigate the bond behavior of the EB composite fabric. To this aim, a series of concrete notched beams are prepared with the EB composite fabric applied onto the bottom face of the specimen and tested under three-point-bending. Results from the DIC measurements in two regions of interest of the beam, namely in the concrete area above the notch and at the FRCM-concrete interface, are then discussed and critically interpreted based on the observed macroscopic mechanical behavior and corresponding force-displacement curves. These outcomes are useful to identify the attainment of the concrete material strength, as well as the failure mode of the beam, and to investigate the stress transfer propagation at the FRCM-concrete interface. Moreover, DIC investigation may be potentially useful to anticipate the incipient crack development.

This study is intended to improve visual bridge damage measurement through image analysis of drone inspection images optimized in terms of quality. To optimize the drone-captured image quality, image properties, including brightness, contrast, and sharpness, are investigated and adjusted for this study. As part of the optimized damage measurement, visual inspections are initially conducted using DJI Matrice 210 for a multi-span timber bridge located in Fall River, South Dakota in the United States. To detect the entirety of any damage during the inspections, the drone records three videos of the bridge from which 15,996 images are extracted. From the drone-aided bridge inspection, several different types of damage, encompassing weathering, paint failure, crack, and split, are observed. Among all the images, the high-quality inspection images with distinct damage on certain bridge components such as timber girders are sorted out manually. Then, the various image properties for the high-quality inspection images are iteratively adjusted to optimize the images’ quality. Measurement of the individual damage detected during the inspection are completed with the optimized quality images. Particularly, the major challenge to identify the damaged area on the weathered timber bridge is resolved, as the edge domains of the damage are well-defined visualized and observed efficiently after the optimization of image quality. Key results indicate that the weathering area identified through the optimized image was measured to be 1.16 m2 but the identified area using the unfiltered image was 1.33 m2, showing a difference of 12.43%.

Previous research on carbon nanofiber aggregates (CNFAs) have shown stable and reliable results for stress sensing with alternating current. The CNFAs necessarily contain carbon nanofibers (CNFs), responsible for the conductivity of this sensor. In addition to the other ingredients present in a CNFA, two 23 gauge welded galvanized steel wire meshes are embedded inside the CNFA cube to create a bridge between the external electrical impedance measuring device and the electrical impedance produced in the CNFA cube. Traditionally, CNFs were mixed in the cement mortar during CNFA casting. However, in this research, CNFAs are cast with CNFs glued to the wire meshes using conductive adhesive to improve the sensitivity of the cement-based sensor. CNFAs were tested under progressive monotonic uniaxial compression for the sweep-frequency test at the frequency range of 500 Hz to 300 kHz. A comparative study on the results from the two sets of CNFA tests shows that sensitivity can be improved when CNFS are glued on wire meshes with conductive adhesive. This paper includes the study of CNFA casting with new technique that increases the accuracy and reliability of sensors in structural health monitoring.

In the last decades, the interest on fiber reinforced polymers (FRPs) has increased due to their mechanical properties and weight saving potential. This has led to the development of novel inspection techniques, being the failure modes of composite structures complex to identify. In this regard, carbon nanotube (CNT) have been widely used for Structure Health Monitoring (SHM) purpose, thanks to their excellent electrical properties and piezoresistive behavior. In this framework, some studies in the literature prove the sensitivity of the CNT percolating network to potential impact damage, rarely exploiting the real-time potential for signal acquisition and never attempting the monitoring during compression after impact (CAI) tests. In this study, the signals from multiple channels on a single GFRP plate specimen doped with multi-walled CNTs have been acquired in real-time and then correlated with simultaneous measures of displacement and forces, aimed at identifying the occurrence of impact damage. In a second phase, the signals from the CNT percolating network have been acquired during CAI tests, identifying precursors of specimen failure.

Minimising the change in guided wave caused by temperature variation rather than occurrence of damage is a key step to reliable damage diagnosis using guided wave based structural health monitoring. To account for the influence of temperature variation on guided wave signals in complex structures, a large amount of baseline measurements need to be collected over a required temperature range to serve as a library. The establishment of baseline library is repeated for each monitored structures, which, if not impossible, is highly impractical. This paper presents a data-driven temperature baseline reconstruction approach that is applicable for various structures made from the same material. The influence of temperature on the amplitude and phase of guided wave measurements are experimentally quantified as dimensionless compensation factors. The derived compensation factors are used for reconstructing baseline signals at various temperatures for guided wave signals. The proposed temperature compensation method is implemented in detecting and localising barely visible impact damage (BVID) in two anisotropic CFRP composite structures, a simple flat plate and a stiffened panel, under different temperature conditions. By compensating the influence of temperature, BVID is successfully detected and located for temperature difference between baseline and current signals up to 25 $$^\circ $$ ∘ C and $$20\,^\circ $$ 20 ∘ C, respectively.

In recent years there has been a substantial increase in the rate-of-penetration in drilling operations with the aim of reducing costs and reaching the reservoir in shorter times. This improvement comes as a result of the greater capabilities of land rigs which are now able to push larger amounts of energy into the drilling system. Consequently, the stress on downhole tools has increased and fatigue failures are becoming more common. Typical failures include cracks in the weaker spots of the downhole tools. In this work we demonstrate that downhole vibration measurements can be used for prognostic health monitoring and to track the structural integrity of the tools. We show evidence that a fundamental downhole frequency mode drops 30% during drilling operations as a crack propagates in the tool. A structural model of the drilling tool is presented, illustrating the relationship between the natural frequency and the localized crack stiffness. By monitoring this frequency shift in real-time, catastrophic and costly failures can be avoided.

In this paper, the parameters of numerical models of beam-like structures are estimated from scalar indicators derived from ambient vibration measurements. Parameters are estimated through an optimization process, in which indicators are taken as objective function. This approach is developed for an indicator, chosen from literature, derived from the reference-based covariance-driven stochastic subspace analysis. The reliability of the parametric identification is further estimated: the method proposed is numerically tested and then applied to a laboratory steel beam. Both simulated and measured vibration data are used to validate the practicability and accuracy of the approach.

In offshore reinforced concrete (RC) structures, how to monitor early occurrences and developments of rebar corrosion is of great significance. However, the rebar corrosion in RC structures is localized and developed along interfaces, having great challenges to evaluations of rebar corrosion levels. In this paper, the theoretical analysis, numerical calculation, and experiment validation are used to study the rebar local corrosion monitoring and evaluation of RC structures with piezoelectric ultrasonic guided waves (UGWs). A reasonable selection of the UGW excitation and reception method and corresponding experimental setup are studied. Frequency dispersion curves of the selected UGWs under different corrosion conditions are obtained by analyzing the wave dispersion and multimodal characteristics. Based on energy values and fractal dimension characteristic values of echo signals for different corrosion levels, a rebar corrosion evaluation index is proposed, and a corresponding evaluation algorithm is established. The effectiveness of the proposed algorithm is verified by a rebar corrosion monitoring test based on the accelerated corrosion and guided wave technologies. A fitting relationship between corrosion levels (length and thickness) and basic characteristics of sensing signals is established. A corrosion evaluation method is established based on the corrosion index and algorithm. The results show that rebar corrosions have a sensitive effect on the energy and fractal characteristics of longitudinal UGWs. The larger corrosion length and the thicker corrosion layer result in the smaller energy value of echo signal and the smaller fractal characteristic value, and the larger corrosion index value.

In this paper results of application of electromechanical impedance (EMI) method for damage assessment in composite panel is presented. Artificial damage simulated by additional mass and real impact damage are investigated. Large number of measurements is analyzed and possibility of utilization of machine learning approach for clustering the damage cases is validated. Clustering is based on principal components obtained based on conductance spectra. Machine learning methods based on K-nearest neighbor, neural networks and K-means are utilized. Moreover, popular damage indices RMSD and CCD were also utilized.

RTM (Resin Transfer Molding) is a well-known closed-mold injection process widely used in manufacturing aeronautical composite parts in low and medium volume production. Resin viscosity and temperature, pressure gradient between ports and inlet and outlet positions in the mold are some of the key parameters of such process as well as fabric material properties, and more in particular, fabric permeability. This last parameter plays an essential role in the infusion process by affecting the advance of the resin flow and it has high influence on possible sources of voids and defects the same as on porosity. This paper describes the technology and feasibility of the Rayleigh distributed optical fiber technique to visually inspect the resin flow inside the mold in two RTM trials. In both of them, resin flow was monitored by studying the variation of the optical fiber backscattering as resin was impregnating the fabrics and also by recording it with a video camera. Results obtained in both trials were analyzed and compared one each other. This online inspection technique presented in this work might be used for detecting irregularities in the resin flow such as either too low/fast speed or non- impregnated areas and moreover, it opens the possibility of modifying in-time injection parameters in RTM processes which will allow correcting possible defects before the part is finished.

There is a growing need to automatically detect and monitor functional changes such as fatigue, wear, damage, or age in the structural components. It is even more attractive if the sensors are embedded in the structure for in-situ structural health monitoring (SHM). In this work, we developed Mechanoluminescence-based sensors for damage sensing and SHM. A flexible, sensitive, and self-powered pressure sensor was developed by integrating a mechanoluminescent device with a light-absorbing layer of the perovskite material. Understanding the behavior of the device as an embedded sensor is necessary for its application in continuous in-situ SHM of multifunctional composite materials. We showed a successful encapsulation process and the embedment of this sensing device into a glass fiber composite. Further investigations were performed to confirm the capability of the embedded sensor to detect the damages in the structures. The embedded sensor could be an effective method for real-time in-situ SHM.

Additive manufacturing of large-scale metal components has a great potential to be applied in the field of equipment manufacturing, repairing, and remanufacturing. The wire and arc additive manufacturing (WAAM) technology has become the preferred additive manufacturing technology for the production of large parts due to its higher deposition rate and better flexibility. The WAAM procedure may lead to a significant periodicity and inhomogeneity in material deposition layers, affecting mechanical properties of materials. The in-situ fatigue testing of Ti-6Al-4V specimens made by WAAM is performed to study fatigue performance based on in-situ observations with scanning electron microscope (SEM). The crack initiation and growth behaviors can be monitored during the fatigue testing of the specimens. The microstructure effect on crack growth rate and path are observed. The uncertainty of fatigue crack growth rate due to these factors is quantified using the experimental data with Paris’ model.

We report on a simple way to enhance the photoresponse and efficiency of inorganic cesium lead halide (CsPbIBr2) perovskite for use as a light-absorbing layer in photodetectors integrated with a mechanoluminescent (ML) or triboluminescence (TL) materials for pressure sensing applications. Herein, we proposed to integrate a thermal and moisture stable inorganic cesium lead halide-based CsPbIBr2 perovskite with the TL materials to develop a novel pressure sensor for real-time and in-situ structural health monitoring (SHM) of aerospace vehicles’ fuselage, automobiles and structures. However, the inorganic CsPbIBr2 perovskite layer fabricated using a one-step spin-coating method is usually composed of small grain size with a large number of grain boundaries and compositional defects. Therefore, we employed a metal doping approach to enhance the CsPbIBr2 perovskite film quality. By introducing a small amount of silver iodide (AgI), the photoresponse, responsivity, and response time of the detector were enhanced. This work offers a promising approach for developing an integrated ML pressure sensor with high-quality polycrystalline perovskite for SHM.

The study presents a probabilistic modeling of the fatigue crack growth prediction of the pipeline steel weldments in nuclear power plants in the context of an integrated structural health monitoring setting. Fatigue testing of the crack growth in the fusion line region of the steel weldments is made using compact-tension specimens. In particular, the uncertainty of the crack growth due to different crack plane orientations is investigated in details. A total of six orientations of the specimens are manufactured and tested according to the ASTM standards to obtain the fatigue crack growth data. The Bayesian method is used to identify the probability density function of the parameters of the Paris’ fatigue crack growth model. Using the concept of damage tolerance, the reliability model of the pipeline weldments given the minimal detectable internal flaws of the ultrasonic nondestructive evaluations can be established. The time-dependent reliability of the pipeline weldments is obtained using the efficient first-order reliability method. Results indicate the uncertainty of the orientations of the flaws plays an important role in the overall reliability of the pipeline weldments.

In order to ensure the safety of railway operation, it is urgent to strengthen the detection and protection of railway perimeter safety. This paper proposes a method for detecting foreign body intrusion in railway perimeter based on double recognition. The Gaussian Mixture Model (GMM) is used to process the video image of the railway scene, and the foreign objects are pre-screened, and the foreign object existence frames are extracted, and then the YOLOv3 algorithm is used to perform secondary detection and recognition on the foreign object existence frames. This method can improve the accuracy of target recognition, reduce the false alarm rate and false alarm rate of foreign object invasion in railway scenes, and occupy less on-site computing resources, which is suitable for on-site requirements. The results show that, compared with the GMM, the false negative rate of the algorithm in this paper is lower, and the algorithm is more suitable for railway site requirements than the deep learning algorithm.

Ultrasonic guide waves (GWs) have increasingly been adopted in Structural Health Monitoring (SHM) of plate-like structures because of their versatility. To operate, SHM based on GWs generally adopt arrays of piezoelectric transducers and imaging techniques for damage detection, localization and quantification. Among the different ultrasonic imaging techniques, Delay-and-Sum (DAS) beamforming applied to Lamb waves is one of the most exploited methods due to the low noise sensitivity and the advantageous trade-off between the image resolution and the number of required sensors. However, DAS shows a limited imaging resolution and contrast which is emphasized as the number of sensors decrease. To tackle these limitations, the scientific literature offers an alternative nonlinear beamforming algorithm called Delay-Multiply-and-Sum (DMAS), which has been successfully applied in RADAR systems and medical ultrasound beamforming.In this work, the DMAS algorithm is applied on guided waves signals to map damages in a composite plate. To the best of author’s knowledge, the DMAS has never been adopted before in the SHM context.In particular, the freely available Guided Waves dataset Open Guided Waves ( http://openguidedwaves.de/ ) that collects piezoelectric actuated and received guided waves signals travelling through a quasi-isotropic composite plate in different damage conditions has been exploited. The DMAS performance has been investigated and compared with the conventional DAS, showing improved imaging resolution and defect localization capabilities.

Structural health monitoring methods based on the generation and detection of highly nonlinear solitary waves have emerged as a cost-effective technique for a variety of structures and materials. Outlier analysis is a statistic tool able to identify anomalies in data that diverge from a set of baseline data. In this paper the use of outlier analysis in terms of discordancy test and Mahalanobis squared distance was explored to enhance the damage detection capability based on the propagation and detection of highly nonlinear solitary waves. An experiment was performed to demonstrate the procedure. A thick steel plate was monitored with a solitary wave transducer placed above the plate, and damage was simulated in terms of a foreign object attached to the bottom of the plate. Three different masses located at an increasing distance from the transducer were considered. A few features were extracted from the experimental time waveforms, and then fed to a univariate and a multivariate analysis that compared the testing data to a set of baseline data. The experimental results show that the outlier analysis significantly enhances the ability to detect damage.

Additive manufacturing (AM) is a common name for a group of techniques that are applied for constructing three-dimensional objects in a layer-by-layer process. The main advantages of such methods are variety of materials (polymers, metals, ceramics) and possibility of manufacturing elements with complex shapes. Therefore, such techniques have been already adopted for rapid prototyping results in shortening delay between design concept and final product.One of the AM methods is multi-jet printing, which offers high accuracy of printed polymeric elements, that can be applied in many industrial branches, e.g. energetic. Safety requirements related to exploitation of structures results in development of structural health monitoring (SHM) methods based on fibre optic sensors. One of the sensor types are fibre Bragg grating (FBG) sensors. Their advantages (small dimensions, multiplexing capabilities) allow them to be embedded into AM structure during manufacturing process.The goal of the paper is to analyse temperature influence on FBG sensors embedded into an AM polymeric material. The analyses will be concerned on both spectrum reflected from the sensor and strain determined using on Bragg wavelength change.

Fibre reinforced polymers are commonly used in many industrial branches. The continuous technical progress in the applied science and technology requires more and more advanced materials. The structural damage can occur due to many factors which are difficult to predict in advance.Safety and reliability requirements results in development of a variety of structural health monitoring (SHM) systems and non-destructive testing (NDT) techniques. In the paper the comparative studies of two non-destructive testing (NDT) methods (infrared thermography and THz spectroscopy) are presented. NDT techniques that can be applied for evaluation of internal structure of composite materials are infrared thermography (pulse and/ or vibrothermography) and THz spectroscopy. Both methods can be used for identification of material structural disintegrations. Infrared thermography allows to observe changes of temperature field distribution, while THz spectroscopy allows to observe changes of absorption coefficient, refractive index or scattering of THz waves propagating throughout analysed material.The goal of the paper is to study the sensitivity and applicability limitations of proposed methods with application to fibre reinforced polymers under simultaneous temperature (form negative to elevate) and relative humidity influence.

Structural Health Monitoring of aeronautic composite structures through Lamb waves can advantageously exploit the fact that Lamb wave damage interaction is nonlinear. However, one difficulty in this context is to be able to distinguish between nonlinearities due to the propagation (i.e. material or geometrical nonlinearities) and those due to the damage itself that are of main interest here. This work proposes to use the Green-Volterra formalism to build up a model for Lamb Wave propagation and damage interaction that is complex enough to represent both types of nonlinearities, and simple enough to be used for simulation and estimation purposes. This approach is presented for the low frequency S0 mode nonlinear propagation in a damaged beam. An analytical model of the nonlinear wave propagation is first derived, where the damage is represented with a polynomial stiffness characteristic acting via boundary conditions. This model is then used to derive the Green-Volterra series describing the nonlinear input-output relationship of the system. A modal decomposition of the Green-Volterra series is also provided. Simulations are presented, and the proposed approach is successfully compared to state-of-the-art methods based on finite-elements models.

We present a method for damage detection and localisation based on multiple-input finite impulse responses. A validation is carried out using measurement data obtained for a girder mast structure.The damage localisation using output-only vibration measurement data of beam-like structures has been a research topic for a long time and many methods have been developed to tackle the problem. However, the identification of finite impulse response filters with multiple inputs has not yet been covered in great detail in the context of these SHM methods.To localise structural damage, first the healthy structure is dynamically excited and finite impulse filters are derived using acceleration sensor data. In this process, we use multiple adjacent sensors as an input to derive the impulse response on a local level in the girder mast structure. The derived filters are applied to obtain an estimation of the transient response for healthy as well as damaged states. Residual signal energies between measured and predicted data are calculated, which increase locally when structural damages occur, enabling the localisation.The outlined damage localisation method, which solely relies on finite impulse responses as an output-only model of the structure, yields promising results in experimental validation.

Numerical simulations of Structural Health Monitoring processes based on wave propagation can be very costly in terms of computation time, especially for complex aeronautic composite structures, and therefore strongly limits the deployment of industrial applications. Metamodels build a relatively simple relationship between inputs and outputs from a set of data and thus can overcome that difficulty. A metamodel based on radial basis functions interpolation is build in order to predict a Lamb Wave measurement on a damaged composite plate equipped by a network of 3 piezoelectric elements. The input parameters describe the position of the damage. This surrogate model is used to predict the measured signals for new damage configurations with a limited computational cost. Moreover, this metamodel is used in a reverse way to solve the inverse problem. A swarm particle optimisation algorithm tries to find the position of a damage from a set of simulated signals. This approach allows us to identify correctly the damage localisation for an unknown configuration, providing therefore a new method for damage localization.

Tensegrity structures form a special class of truss with dedicated cables and bars, that take tension and compression, respectively. To ensure equilibrium, the tensegrity members are required to be prestressed. Over prolonged usage, the cables may lose their prestress while bars may buckle, affecting the structural stiffness as well as its dynamic properties. The stiffness of tensegrities also vary with the load even in the absence of damage. This can potentially mask the effect of damage leading to a false impression of tensegrity health. This poses a major challenge in tensegrity health monitoring especially when the load is stochastic and unknown.Present study develops a vibration based output-only time-domain approach for monitoring the health of any tensegrity in the presence of uncertainties due to ambient force and measurement noise. An Interacting Particle Ensemble Kalman Filter (IPEnKF) has been used that can efficiently monitor tensegrity health from contaminated response data. IPEnKF combines a bank of Ensemble Kalman Filters to estimate response states while running within a Particle Filter envelop that estimates a set of location based health parameters. Further to make damage detection cheaper, strain responses are used as measurements. The efficiency of the proposed methodology has been demonstrated using numerical experiments performed on a simplex tensegrity.

The application of lithium-ion batteries is widely spread nowadays. They can be found especially in consumer electronics such as mobile phones and notebooks, as well as in the rapidly growing field of electric cars. Highly available power and energy density of a lithium-ion battery over its lifetime are crucial for their effective operation. A replacement of the battery is recommended if the capacity is reduced by aging to below 80% of the initial capacity. For the estimation of the remaining cell-life, a precise estimation of the state of charge (SoC) and state of health (SoH) is necessary. This cannot be assured by common battery management systems, which only rely on current counting and cell voltage measurements. This work presents the potential to determine the SoC and SoH of lithium-ion cells with elastic guided waves. As the propagation of guided waves is dependent on the density and elastic modulus of a media, a correlation between the state of charge and the measured amplitude and phase of the wave can be shown, as the porosity of the anode will change during charge and discharge. Experimental studies in a pitch-catch arrangement of piezoelectric transducers were conducted on a representative pouch cell. The attached transducer network allows the examination of multiple travel paths through the battery.

Considering technically complex systems, the evaluation of situations or conditions is a challenging task. To ensure a high accuracy assignments from different classifiers can be fused. To define requirements for a good fusion performance and to evaluate the potential for higher accuracy, in this paper the idea of a fictional classifier is introduced. The precision values of one classifier denoted as fictional classifier are varied to demonstrate the influence on fused accuracy. From the results of this contribution some important challenges can be solved: Can the performance of one individual classifier improve the overall accuracy? Can the performance of the fused results be improved by changing performance measures of the fictional classifier? This allows the establishment of a supervised strategy to adapt precision values to get better fusion results. For illustrating the effectsfour benchmark examples are used. The introduced methods are applied to fault diagnosis of hot rolling mills. The results show that using a fictional classifier the overall accuracy can be outperformed depending on data sets.

The Internet of Things (IoT) is shaping the concept of the modern intelligent built environment. The latest developments in IoT have led to secure, energy efficient systems enabling low-cost real-time analytics. In the Vertical Transportation (VT) technologies developed by the lift industry real-time analytics are facilitating predictive maintenance which in turn decreases operational and downtime costs. Data driven predictive maintenance does not always reach an optimal performance because the quality and quantity of the data matters. Fault classification and the estimation of the remaining useful life (RUL) requires a deep understanding of failure modes and component degradation. In lift systems, most of the malfunctions are due to faults developed by the automatic power operated door systems. The most widespread Structural Health Monitoring (SHM) sensor technology used in monitoring the door mechanisms are acoustic and vibration sensors. In this paper, an automatic fault detection system using Artificial Neural Networks (ANN) is implemented using vibration signal features. Obtained results reveal that the fault classification performance is high (>70%) under low noise environmental conditions.

In an earlier research, experimental evidence was given on the ability to use Piezo Wafer Active Sensors and acousto-ultrasonics to monitor the accumulation of fatigue damage in a thick composite structure. As a next step, numerical models are investigated as they aid in the further understanding of the governing phenomena and a quantification of the accumulated damage. However, they suffer from high computational demands, due to a high mesh density, the stochastic nature of crack initiation and the combination of initiation and propagation of cracks.The Polynomial Chaos Expansion (PCE) method is employed to efficiently make meta models and, with these models, account for the stochastic behaviour of crack initiation and formation of delaminations. The meta models thus allow predicting the overall effect of damage accumulations within certain bounds of uncertainty. This aids in the quantification of damage accumulation, hence allowing for a damage severity estimation based on the experimental results.The input for the PCE method is a 2D Finite Element (FE) model. Cracks and delaminations are generated using Random Variables (RV) describing the geometrical position and length and orientation. Moreover, the number of cracks and delaminations is randomized as well. The necessary remeshing is done automatically, allowing for a completely automated simulation for a large number of FE simulations to feed the PCE model.Several Quantities of Interests (QoI) are defined and tested against their sensitivity to the increasing amount of damage accumulation. A global sensitivity analysis is used to identify the importance of each of the Random Variables. Random variables with a low sensitivity can be eliminated from the analysis, improving the efficiency.

This paper presents an experimental and numerical study on multiple damage localization in a beam. The modal rotations of an aluminum beam were measured with shearography and post-processed to obtain the modal curvatures. The modal curvatures, which are computed by finite differences, are used as damage indicators. In most approaches available in the literature, the modal curvatures are defined from the modal displacements, requiring the computation of the second order derivate. In the present approach, since the modal rotations are available, the curvatures are obtained by computing only the first order derivative, reducing the propagation of measurement errors. Optimal samplings for both the forward and the central finite difference schemes, the latter with three and five points formulas, are derived. The results of applying these three finite difference formulas are compared, showing that both central finite differences allow for a better representation of the experimental modal curvature. Therefore, the perturbations on the modal curvatures are better identified, thus clearly indicating the damage presence.

This contribution presents and tests experimentally a nonparametric approach for indirect identification of 2D paths of moving loads, based on the recorded mechanical response of the loaded structure. This is an inverse problem of load identification. The method to be proposed is based on multicriterial optimization with two complementary criteria. The first criterion is purely mechanical, and it quantifies the misfit between the recorded mechanical response of the structure and its predicted response under a given trajectory. The second criterion is geometric: it represents the heuristic knowledge about the expected geometric regularity characteristics of the load paths (such as related to linear and angular velocity), and in fact it can be considered to be a regularizing criterion. A multicriterial genetic search is used to determine and advance the Pareto front, which helps to strike the balance between the response fit and the geometric regularity of the path. The proposed approach is tested in an experimental laboratory setup of a plate loaded by a line-follower robot and instrumented with a limited number of strain gauges.

Electromechanical Impedance (EMI) method employs high frequencies range in assessing the local structural damage, using various damage metrics. These dam-age metrics are used as a tool to separate quantitatively EMI spectra into classes of the damage presence level and location in the structure. This paper describes the EMI based damage quantification using two piezoelectric transducers on the steel beam structure. The series and parallel combinations were studied and compared with the output of single sensors. The theoretical approach for simulating the serial and parallel connections was proposed and tested. The performance comparison was done in the selected frequency range: 600–800 kHz. The advantage of using the parallel connection was shown on the considered example of the steel beam with and additional mass. The mass was successfully detected and the time needed for measurements was reduced. It was also shown that the simplified model of parallel connection gives the comparison result to the real connection of the two transducers.

A study on the post-processing of modal displacements and modal rotations with the Haar wavelet is presented in this paper. A free-free aluminum beam is chosen as a structure to be analyzed and its modal displacements and rotations in both undamaged and multiple damaged states are computed with the finite element method. Three sets of damage indicators are proposed: (1) wavelet transform of damaged modal displacements or modal rotations, (2) wavelet transform of differences in modal displacements or modal rotations, and (3) ratio of wavelet transform of differences in modal displacements or modal rotations to wavelet transform of undamaged modal displacements or modal rotations. The study shows that post-processing modal displacements with the Haar wavelet leads to poor damage identifications. On the other hand, if one applies the same type of wavelet to modal rotations, the damage indicators (2) and (3) clearly pinpoints the location of the damage. Although the set of damage indicators (1) and (2) are prone to the boundary effect, where high values of wavelets coefficients are present, the third set of damage indicators does not present this problem. Furthermore, for large values of damage, this damage indicators shows the presence of damage in all scales of the wavelet scalogram. The present study clearly shows that by post-processing modal rotations with the simplest wavelet one can obtain reliable identifications of multiple damage.

In view of diversified applications in aerospace, civil, and mechanical engineering, Lamb wave based NDE of composite laminates has been of continued interest. In particular, the techniques based on the nonlinear wave damage interactions have attracted significant attention in the recent past. A delamination defect with contacting interfaces (referred to as a breathing delamination) is a potential source of nonlinearity because of the bilinear stiffness characteristics instigating through an intermittent contact. Past investigations have dealt predominantly with a single frequency excitation leading to the generation of higher harmonics that can be utilized as a viable damage indicator. In this work, we investigate numerically the nonlinear interactions of a dual frequency Lamb wave signal with breathing delamination in a composite plate. To this end, an eight layer GFRP laminate is considered in the analysis. Explicit dynamic simulations are performed using commercial finite element software ANSYS. It is demonstrated that the nonlinear wave-damage interactions lead to frequency side-bands in the response spectrum occurring at the algebraic combinations of the two constituent frequencies. A modulation parameter is introduced for quantifying the strength of the combination harmonics relative to the fundamental harmonics. Further, a thorough parametric study is performed for assessing the impact of ply orientations and inter-laminar position of the delamination on the proposed modulation parameter. This investigation can find its potential use in devising damage localization strategies for laminated composites based on the nonlinear wave-damage interactions.

In the last decades, structural health monitoring (SHM), and, in particular, early damage detection methodologies, have emerged as an important tool to assist in the maintenance and management of infrastructures. In this context, this work presents a methodology for damage detection in full-scale bridges based on moving-loads responses. The Luiz I bridge, an outstanding centenary steel double-deck arch bridge, was selected as a case study. The methodology consists in building time-series of vehicle-influence lines of the strains observed in the selected cross-sections and processing data by using moving principal component analysis (MPCA). Firstly, the effectiveness of the approach is assessed on numerically-simulated data, to show, on the one hand, the stability of the approach under undamaged conditions, and, on the other, the ability of the approach to highlight changes in the structural condition. Finally, the methodology is applied on field data collected for a 3-months period.