Showing papers in "Structural Health Monitoring-an International Journal in 2022"

Statistics-based baseline-free approach for rapid inspection of delamination in composite structures using ultrasonic guided waves

Abstract: Delamination in composite structures is characterized by a resonant cavity wherein a fraction of an ultrasonic guided wave may be trapped. Based on this wave trapping phenomenon, we propose a baseline-free statistical approach for the identification and localization of delamination using sparse sampling and density-based spatial clustering of applications with noise (DBSCAN) technique. The proposed technique can be deployed for rapid inspection with minimal human intervention. The Performance of the proposed technique in terms of its ability to determine the precise location of such defects is quantified through the probability of detection measurements. The robustness of the proposed technique is tested through extensive simulations consisting of different random locations of defects on flat plate structures with different sizes and orientation as well as different values of signal to noise ratio of the simulated data. The simulation results are also validated using experimental data and the results are found to be in good agreement.

Crack detection of concrete structures using deep convolutional neural networks optimized by enhanced chicken swarm algorithm

Abstract: With the rapid increase of ageing infrastructures worldwide, effective and robust inspection techniques are highly demanding to evaluate structural conditions and residual lifetime. The damages on structural surfaces, for example, spalling, crack, rebar buckling and exposure, are important indicators to assess the structural condition. In fact, several state-of-the-art automated inspection techniques using these indicators have been developed to reduce human-conducted onsite inspection activities. However, the efficiency of these techniques is still required to be improved in terms of accuracy and computational cost. In this study, a vision-based crack diagnosis method is developed using deep convolutional neural network (DCNN) and enhanced chicken swarm algorithm (ECSA). A DCNN model is designed with a deep architecture, consisting of six convolutional layers, two pooling layers and three fully connected layers. To enhance the generalisation capacity of trained model, ECSA is introduced to optimize meta-parameters of the DCNN model. The model is trained and tested using image patches cropped from raw images obtained from damaged concrete samples. Finally, a comparative study on different crack detection techniques is conducted to evaluate performance of the proposed method via a group of statistical evaluation indicators.

Bayesian dynamic regression for reconstructing missing data in structural health monitoring

Abstract: Massive data that provide valuable information regarding the structural behavior are continuously collected by the structural health monitoring (SHM) system. The quality of monitoring data is directly related to the accuracy of the structural condition assessment and maintenance decisions. Data missing is a common and challenging issue in SHM, compromising the reliability of data-driven methods. Thus, the accurate reconstruction of missing SHM data is an essential step for the reliable evaluation of the structural condition. Data recovery can be considered as a regression task by modeling the correlation among sensors. The Bayesian linear regression (BLR) model has been extensively used in probabilistic regression analysis due to its efficiency and the ability of uncertainty quantification. However, because of the fixed coefficients (refer to a static model) and linear assumption, the BLR model fails to accurately capture the relationship and accommodate the changes in related variables. Given this limitation, this study presents a Bayesian dynamic regression (BDR) method to reconstruct the missing SHM data. The BDR model assumes that the linear form is only locally suitable, and the regression variable varies according to a random walk. In particular, the multivariate BDR model can reconstruct the missing data of different sensors simultaneously. The Kalman filter and expectation maximum (EM) algorithms are employed to estimate the state variables (regressors) and parameters. The feasibility of the multivariate BDR model is demonstrated by utilizing the data from a building model and a long-span cable-stayed bridge. The results show that the multivariate BDR model exhibits excellent performance to rebuild the missing data in terms of both computational efficiency and accuracy. Compared to the standard BLR and linear BDR models, the quadratic BDR model owns better reconstruction accuracy.

Looseness monitoring of multiple M1 bolt joints using multivariate intrinsic multiscale entropy analysis and Lorentz signal-enhanced piezoelectric active sensing

Abstract: Bolts are widely used in the fields of mechanical, civil, and aerospace engineering. The condition of bolt joints has a significant impact on the safe and reliable operation of the whole equipment. The failure of bolt joints monitoring leads to severe accidents or even casualties. This paper proposes a novel bolt joints monitoring method using multivariate intrinsic multiscale entropy (MIME) analysis and Lorentz signal-enhanced piezoelectric active sensing. Lorentz signal is used as excitation signal in piezoelectric active sensing to expose nonlinear dynamical characteristics of the bolt joints. Multivariate variational mode decomposition (MVMD) is employed to decompose multiple components of the collected Lorentz signal into multivariate band-limited intrinsic mode functions (BLIMFs). Afterward, improved multiscale sample entropy (IMSE) values of each channel’s BLIMFs are computed to measure its irregularity and complexity. IMSE values are taken as quantitative features, reflecting dynamical characteristics of bolt joints. Further, the constructed 3-layer feature matrices are adopted as the input of the convolutional neural network (CNN) to achieve accurate bolt joint monitoring. The multiple M1 bolt joints are used during the experiment to verify the effectiveness and superiority of the proposed approach. The results demonstrate the proposed novel approach is promising in bolt joints monitoring.

Three decades of statistical pattern recognition paradigm for SHM of bridges

Abstract: Bridges play a crucial role in modern societies, regardless of their culture, geographical location, or economic development. The safest, economical, and most resilient bridges are those that are well managed and maintained. In the last three decades, structural health monitoring (SHM) has been a promising tool in management activities of bridges as potentially it permits one to perform condition assessment to reduce uncertainty in the planning and designing of maintenance activities as well as to increase the service performance and safety of operation. The general idea has been the transformation of massive data obtained from monitoring systems and numerical models into meaningful information. To deal with large amounts of data and perform the damage identification automatically, SHM has been cast in the context of the statistical pattern recognition (SPR) paradigm, where machine learning plays an important role. Meanwhile, recent technologies have unveiled alternative sensing opportunities and new perspectives to manage and observe the response of bridges, but it is widely recognized that bridge SHM is not yet fully capable of producing reliable global information on the presence of damage. While there have been multiple review studies published on SHM and vibration-based structural damage detection for wider scopes, there have not been so many reviews on SHM of bridges in the context of the SPR paradigm. Besides, some of those reviews become obsolete quite fast, and they are usually biased towards applications falling outside of bridge engineering. Therefore, the main goal of this article is to summarize the concept of SHM and point out key developments in research and applications of the SPR paradigm observed in bridges in the last three decades, including developments in sensing technology and data analysis, and to identify current and future trends to promote more coordinated and interdisciplinary research in the SHM of bridges.

Monitoring deformations of infrastructure networks: A fully automated GIS integration and analysis of InSAR time-series

Abstract: Ageing stock and extreme weather events pose a threat to the safety of infrastructure networks. In most countries, funding allocated to infrastructure management is insufficient to perform systematic inspections over large transport networks. As a result, early signs of distress can develop unnoticed, potentially leading to catastrophic structural failures. Over the past 20 years, a wealth of literature has demonstrated the capability of satellite-based Synthetic Aperture Radar Interferometry (InSAR) to accurately detect surface deformations of different types of assets. Thanks to the high accuracy and spatial density of measurements, and a short revisit time, space-borne remote-sensing techniques have the potential to provide a cost-effective and near real-time monitoring tool. Whilst InSAR techniques offer an effective approach for structural health monitoring, they also provide a large amount of data. For civil engineering procedures, these need to be analysed in combination with large infrastructure inventories. Over a regional scale, the manual extraction of InSAR-derived displacements from individual assets is extremely time-consuming and an automated integration of the two datasets is essential to effectively assess infrastructure systems. This paper presents a new methodology based on the fully automated integration of InSAR-based measurements and Geographic Information System-infrastructure inventories to detect potential warnings over extensive transport networks. A Sentinel dataset from 2016 to 2019 is used to analyse the Los Angeles highway and freeway network, while the Italian motorway network is evaluated by using open access ERS/Envisat datasets between 1992 and 2010, COSMO-SkyMed datasets between 2008 and 2014 and Sentinel datasets between 2014 and 2020. To demonstrate the flexibility of the proposed methodology to different SAR sensors and infrastructure classes, the analysis of bridges and viaducts in the two test areas is also performed. The outcomes highlight the potential of the proposed methodology to be integrated into structural health monitoring systems and improve current procedures for transport network management.

Non-parametric empirical machine learning for short-term and long-term structural health monitoring

Abstract: Early damage detection is an initial step of structural health monitoring. Thanks to recent advances in sensing technology, the application of data-driven methods based on the concept of machine learning has significantly increased among civil engineers and researchers. On this basis, this article proposes a novel non-parametric anomaly detection method in an unsupervised learning manner via the theory of empirical machine learning. The main objective of this method is to define a new damage index by using some empirical measure and the concept of minimum distance value. For this reason, an empirical local density is initially computed for each feature and then multiplied by the minimum distance of that feature to derive a new damage index for decision-making. The minimum distance is obtained by calculating the distances between each feature and training samples and finding the minimum quantity. The major contributions of this research contain developing a novel non-parametric algorithm for decision-making under high-dimensional and low-dimensional features and proposing a new damage index. To detect early damage, a threshold boundary is computed by using the extreme value theory, generalized Pareto distribution, and peak-over-threshold approach. Dynamic and statistical features of two full-scale bridges are used to verify the effectiveness and reliability of the proposed non-parametric anomaly detection. In order to further demonstrate its accuracy and proper performance, it is compared with some classical and recently published anomaly detection techniques. Results show that the proposed non-parametric method can effectively discriminate a damaged state from its undamaged condition with high damage detectability and inconsiderable false positive and false negative errors. This method also outperforms the anomaly detection techniques considered in the comparative studies.

Robust fault diagnosis of rolling bearing via phase space reconstruction of intrinsic mode functions and neural network under various operating conditions

Abstract: Rolling bearings are important components in mechanical, civil, and aerospace engineering. The practical working conditions of rolling bearings are complex; hence, fault diagnosis of rolling bearings under various operating conditions is very challenging. This paper proposes a novel approach to fault diagnosis of rotary machinery using phase space reconstruction (PSR) of intrinsic mode functions (IMFs) and neural network under various operating conditions. Complete ensemble empirical mode decomposition with adaptive noise (CEEMDAN) is employed to decompose vibration signal of rotary component into IMFs denoting high-to-low instantaneous frequencies adaptively. PSR constructs one-dimensional IMFs to high-dimensional IMFs, which helps reveal the underlying nonlinear geometric topology via the reconstructed inherent and hidden dynamical characteristics of the one-dimensional vibration signal. To explore intrinsic dynamical properties, interquartile range (IQR) of Euclidean distance (ED) values of high-dimensional IMFs are extracted as condition indicators and used as input of back propagation (BP) neural network to fulfill fault identification of rolling bearings. The effectiveness and superiority of the proposed approach have been validated by theoretical derivations, numerical simulations and experimental data. The results show that the proposed approach is promising in fault diagnosis of rotary machinery under various operating conditions.

Robust fault diagnosis of rolling bearings via entropy-weighted nuisance attribute projection and neural network under various operating conditions

Abstract: Rolling bearings are crucial components in the fields of mechanical, civil, and aerospace engineering. They sometimes work under various operating conditions, which makes it harder to distinguish faults from normal signals. Nuisance attribute projection (NAP) is a technique that has been widely used in audio and image recognition to eliminate interference information in the extracted feature space. In constructing the weighted matrix of NAP, the setting of the weighted value represents the degree of interference between the feature vectors. The interference is either taken into consideration in whole, or not considered at all, which will inevitably lead to information loss. In our work, an entropy-weighted NAP (EWNAP) is proposed to deal with such “bipolar problem” in constructing the weighted matrix. The eigenvalues of covariance matrix of collected signals contain dynamical information, and the fuzzy entropy is adopted to evaluate the dispersion degree of these eigenvalues. After normalization, these entropy values are used to express the weight relationship in the weighted matrix of EWNAP. The features processed by EWNAP can be used as samples and combined with neural network to achieve fault diagnosis of rolling bearings. Furthermore, a fault diagnosis approach with insufficient data is demonstrated to validate the effectiveness of the proposed scheme. In the case studies, Case Western Reserve University bearing database and data collected from the bearing fault simulation bench are used. These case studies show that the proposed EWNAP alleviates the interference caused by various operating conditions, and the comparative analysis confirms that the proposed method works better than the conventional methods.

Deep learning enhanced principal component analysis for structural health monitoring

Abstract: This paper proposes a Deep Learning Enhanced Principal Component Analysis (PCA) approach for outlier detection to assess the structural condition of bridges. We employ partially explainable autoencoder architecture to replicate and enhance the data compression and reconstruction ability of PCA. The particularity of the method lies in the addition of residual connections to account for nonlinearities. We apply the proposed method to monitoring data obtained from two bridges under real operation conditions and compare the results before and after adding the residual connections. Results show that the addition of residual connections enhances the outlier detection ability of the network, allowing to detect lighter damages.

A novel adaptive bandwidth selection method for Vold–Kalman filtering and its application in wind turbine planetary gearbox diagnostics

Abstract: The planetary gearbox transmission system in wind turbines has complex structures and generally operates under non-stationary conditions. Thus its measured responses are of high complexity and nonlinearity, which brings a great challenge in the development of reliable condition monitoring techniques for the planetary gearbox transmission system. As a prevalent and effective tool for analyzing the non-stationary vibration signal with strong nonlinearity, the Vold–Kalman filtration technique has excellent capabilities of tracking the targeted harmonic components of vibrations, which can significantly benefit planetary gearbox fault diagnostics. However, the tracking accuracy is heavily enslaved to the selection of the rational bandwidth for the Vold–Kalman filter. An inappropriate bandwidth could impair the characteristics of the targeted harmonic responses, and as a consequence, the monitoring process becomes no longer reliable. To address this issue, a novel bandwidth selection methodology for the Vold–Kalman filter is developed in this paper. Through comprehensively depicting the targeted harmonic response using features in multiple domains, the rational bandwidth can be selected for Vold–Kalman filtering, and then, a reliable monitoring process can be ensured. Additionally, a tacho-less speed estimation procedure is utilized in this paper to acquire the instantaneous rotational speed from the vibration signal directly. With the rational bandwidth and the estimated rotational speed, the desired harmonic components of vibrations can be adaptively extracted and tracked through the Vold–Kalman filter with high accuracy, and at the same time, the irrelevant or unwanted components are excluded completely. The effectiveness and superiority of the proposed adaptive Vold–Kalman filtration for wind turbine planetary gearbox diagnostics are demonstrated and validated experimentally.

Temperature compensation to guided wave-based monitoring of bolt loosening using an attention-based multi-task network

Abstract: Online monitoring of bolt torque is critical to ensure the safe operation of bolted structures. Guided waves have been intensively explored for bolt loosening monitoring. Nevertheless, guided waves are excessively sensitive to fluctuation of ambient temperature. As a result of the complexity of wave transmitting across a bolted joint, it is highly challenging to compensate for the effect of temperature. To this end, an attention-based multi-task network is developed towards accurate detection of bolt loosening in multi-bolt connections over a wide range of temperature variation. By integrating improved attention gate modules in a modified U-Net architecture, an attention U-Net is configured for temperature compensation. A two-layer convolutional subnetwork is connected in series behind the attention U-Net to identify bolt loosening. Experimental validation is carried out on a bolt jointed lap plate simulating a real aircraft structure. The results have proved that the developed multi-task network achieves temperature compensation and accurate bolt loosening identification. To further understand the multi-task network, the Integrated Gradients method and a simplified structure of the bolt lap plate are used to interpret the developed network. It is proved that the A0 mode is sensitive to bolt loosening, while the S0 mode is not.

Bayesian dynamic linear model framework for structural health monitoring data forecasting and missing data imputation during typhoon events

Abstract: A Bayesian dynamic linear model (BDLM) framework for data modeling and forecasting is proposed to evaluate the performance of an operational cable-stayed bridge, that is, Ting Kau Bridge in Hong Kong, by using SHM strain field data acquired. One of the major challenges in dealing with the existing in-service bridge under extreme typhoon loads is to forecast structural behavior using the typhoon response exhibiting non-stationarity, large data fluctuations and strong randomness. The first attempt for SHM data modeling during extreme events, that is, typhoons, using BDLM framework, was conducted in this study. The data from multiple sensors are analyzed for one-step, multi-step ahead forecasting and missing data imputation. The overall bridge behavior is incorporated into a forecasting model by superposition of forecasting results of trend (representing the structural baseline response), periodic component (response component evolving regularly over time), and autoregressive component (time-dependent error) through BDLM algorithm. The results demonstrate that the BDLM framework yielded more accurate calculations compared with Gaussian process and Variational Heteroscedasticity Gaussian Process methods with respect to one-step ahead forecasting for strain data under typhoons. Multi-step ahead forecasting was successfully carried out both for non-typhoon and typhoon responses within an acceptable precision range. The correlation between periodic component and temperature was also investigated. Regarding missing data imputation, BLDM algorithm can generate robust results due to making full use of the monitoring data both before and after the missing segments.

Railway defect detection based on track geometry using supervised and unsupervised machine learning

Abstract: Track quality affects passenger comfort and safety. To maintain the quality of the track, track geometry and track component defects are inspected routinely. Track geometry is inspected using a track geometry car (TGC). Measured values are stored in the machine and processed to evaluate the track quality. However, track component defects require more effort to inspect. Track component defects can be inspected manually which is time- and workload-consuming or using sensors installed at additional cost. This study presents an approach using track geometry obtained by a TGC to detect track component defects, namely, rail, switch and crossing, fastener and rail joint defects. Detection models are developed using several supervised machine learnings. The relationships between track component defects are analysed to gain insights using unsupervised machine learnings. From the study, the best model for detecting track component defects using track geometry is a deep neural network with an accuracy of 94.31% followed by a convolutional neural network with an accuracy of 93.77%. For the exploration of insights, k-means clustering is used to cluster the track components defects, and association rules are used to find the relationships between them. Examples of the insights from applying these two techniques are that switch and crossing defects are usually found where the radius of curvature is less than 2000 m and the gradient is positive, the most common defects when the radius of curvature higher than 4000 m are rail defects, or a worn wing rail will be found when the rail section has failed, ties in switches and worn point blades are found with the confidence of 92.17%. The findings of the study can be applied to detect track component defects using track geometry where additional cost is not required and unsupervised machine learning provides the insights that will be beneficial for railway maintenance. The information obtained from machine learning models will be complementary information to support decision making and improve the maintenance efficiency in the railway industry.

A convolutional neural network for pipe crack and leak detection in smart water network

Abstract: The implementation of a smart water network (SWN) is viewed as a strategic approach to address many challenges faced by water utilities, such as pipe leak detection and main break prevention. This paper develops a convolutional neural network (CNN)–based model to classify acoustic wave files collected by the South Australian Water Corporation’s (SA Water’s) SWN over the city of Adelaide. The VGGish model (VGG refers to the team who developed the model—Visual Geometry Group) is selected as a suitable transfer learning model to extract features from wave files. The CNN model classifies an acoustic wave file as an anomaly or other background or environmental noise. Identification of a wave file as an anomaly triggers a Siamese CNN model to determine whether it is related to a regular/irregular scheduled event (for example, irrigation system near public parks or water consumption by large buildings). A field investigation is initiated if a wave file is classified as an anomaly and it is not related to a scheduled event. The developed models have been validated using data that is recorded by SWN in Adelaide. This validation data set comprises 1098 wave files, which are recorded by 34 accelerometers and are associated with 32 known leaks. The validation results shown that accuracy of alarms generated by the developed models is 92.44%. The validations confirm the developed models as an effective tool for water pipeline leak and crack detection, which, in turn, enables proactive management of the pipeline assets.

Engineering deep learning methods on automatic detection of damage in infrastructure due to extreme events

Abstract: This paper presents a few comprehensive experimental studies for automated Structural Damage Detection (SDD) in extreme events using deep learning methods for processing 2D images. In the first study, a 152-layer Residual network (ResNet) is utilized to classify multiple classes in eight SDD tasks, which include identification of scene levels, damage levels, and material types. The proposed ResNet achieved high accuracy for each task while the positions of the damage are not identifiable. In the second study, the existing ResNet and a segmentation network (U-Net) are combined into a new pipeline, cascaded networks, for categorizing and locating structural damage. The results show that the accuracy of damage detection is significantly improved compared to only using a segmentation network. In the third and fourth studies, end-to-end networks are developed and tested as a new solution to directly detect cracks and spalling in the image collections of recent large earthquakes. One of the proposed networks can achieve an accuracy above 67.6% for all tested images at various scales and resolutions, and shows its robustness for these human-free detection tasks. As a preliminary field study, we applied the proposed method to detect damage in a concrete structure that was tested to study its progressive collapse performance. The experiments indicate that these solutions for automatic detection of structural damage using deep learning methods are feasible and promising. The training datasets and codes will be made available for the public upon the publication of this paper.

A Feature Extraction & Selection Benchmark for Structural Health Monitoring

Abstract: There are a large number of time domain, frequency domain and time-frequency signal processing methods available for univariate feature extraction. However, there is no consensus in SHM on which feature, or feature sets, are best suited for the identification, localisation and prognosis of damage. This paper attempts to address this problem by providing a comprehensive benchmark of feature selection & reduction methods applied to an extensive set of univariate features. These univariate features are extracted using multiple statistical, temporal and spectral methods from the benchmark S101 and Z24 bridge datasets. These datasets contain labelled accelerometer recordings from full scale bridges as they are progressively subjected to multiple damage scenarios. To identify the minimal set of features that best distinguishes between the multiple damage states, a supervised machine learning approach is used in combination with multiple feature selection methods. The ability of these reduced feature sets to distinguish between damage states is benchmarked using the prediction performance of the classification models, with the training and test sets obtained through stratified k-fold cross validation. The results obtained show that reduced sets of univariate features, extracted from a single accelerometer sensor, are capable of accurately distinguishing between multiple classes of healthy and damaged states. This work provides a benchmark for SHM practitioners and researchers alike for the choice, comparison and validation of feature extraction and feature selection methods across a wide range of systems.

Comparison of neural networks based on accuracy and robustness in identifying impact location for structural health monitoring applications

Abstract: Structural health monitoring systems must provide accuracy and robustness in predicting the structure’s health using the minimum intervention to ensure commercial viability. Characterization of impact is useful in assessing its severity, deciding if detailed damage analysis is necessary, and re-evaluating the present health of the structure under monitoring with better confidence. In this characterization process, the impact location is significant since some positions within a structure are more sensitive to damage. The inherent noise and uncertainties present in the sensor response pose a substantial hurdle to estimating the external impact correctly. This paper quantitatively compares three of the widely used neural networks, namely, Artificial Neural Network (ANN), Convolutional Neural Network (CNN), and Long Short-Term Memory network (LSTM), to estimate impact location from the lead zirconate titanate (PZT) sensor response. For this purpose, a square aluminum plate of 500 × 500 mm was equipped with four PZT sensors; each placed 100 mm away in both the plate directions from a corner and impact loads were given on a grid covering the whole plate. The PZT responses were used to train the three neural networks under study here, and their estimations were compared based on the Mean Absolute Error (MAE). In addition, increasing Gaussian noise was added to the PZT responses, and the robustness of the three neural networks was monitored. It was found that the ANN gives better accuracy with a Mean Absolute Error of 22 mm compared to Convolutional Neural Network (MAE = 31 mm) and Long Short-Term Memory (MAE = 25 mm). However, CNN is more robust when encountering noise with a 2% reduction in accuracy, while LSTM and ANN lost 7% and 11% accuracy, respectively.

Damage detection on a historic iron bridge using satellite DInSAR data

Abstract: Structural Health Monitoring (SHM) allows tracking the structural behavior in time and support decisions regarding, for instance, the need for maintenance and repair activities. Most traditional SHM systems require sensors that are directly applied to the structure to get insights into the structural performance. Satellite technologies can provide an appealing alternative to traditional SHM. They allow to measure displacements at a large scale and to follow their evolution without the need of directly accessing the structure. Further to this, the possibility to monitor large areas opens new avenues for the development of automatic alert systems able to issue an alarm and early-flag damaged structures. However, displacements of civil structures might also be induced by sources other than damage such as thermal or periodic hydrogeological variations. These can hinder the onset and development of damage or lead to false alarms if such displacements are erroneously interpreted as damages. This paper aims to present a new method for damage detection based on DInSAR measurements, that tackles both aspects providing reliable information about the onset of damage under environmentally changing conditions in a period corresponding to about twice the revisit time of the satellite. A case study is presented to demonstrate the applicability of the proposed method, namely the Palatino bridge in Rome, Italy. The satellite data are acquired by COSMO-SkyMed of the Italian Space Agency and consist of displacements of the observed structure recorded during a period spanning between 2011 and 2019.

A Machine learning-based approach to determining stress in rails

Abstract: Recent advancements in both software and hardware have sparked the use of machine learning (ML) in structural health monitoring (SHM) applications. This paper delves into the use of ML to determine axial stress in continuous welded rails (CWR). The overall proposed SHM strategy consists of monitoring the vibration of CWR and associating their modal characteristics to the rail longitudinal stress using a ML algorithm trained with data generated with a finite element model. In the present study, the feasibility of the proposed strategy was tested on a simple rail segment subjected to mechanical compression. Two algorithms were developed using hyperparameter search optimization techniques to infer the stress from the frequencies of vibration of a few modes of the rail. The training data were generated with a finite element model of a rail segment under varying axial stresses, rail lengths, and boundary conditions at the two ends of the segment. The algorithms were then tested with a second set of data generated numerically and the results of an experiment in which a 2.4-m-long rail was subjected to compressive load and excited with an instrumented hammer. Both tests demonstrated that ML is a viable tool to estimate axial stress in the rail segment provided a sufficient number of modes of vibrations are presented to the learning algorithm. For the future, more experiments are warranted to test the ML against data from real CWR.

Pitting corrosion prediction based on electromechanical impedance and convolutional neural networks

Abstract: Corrosion induced thickness loss in metallic structures is a common and crucial problem in multiple industries. Therefore, it is important to accurately monitor the corrosion amount of the structure. Traditional corrosion monitoring methods are mainly based on electrochemical methods, and most of them are unable to quantify the corrosion amount. In our previous work, a new type of corrosion sensing mechanism based on the electromechanical impedance instrumented circular piezoelectric-metal transducer was proposed, in which the peak frequencies in the conductance signatures decrease linearly with the increase of the corrosion induced thickness loss. However, only the uniform corrosion with even metal thickness decrease was considered in the previous study. In this paper, the capability of the proposed sensing mechanism for the quantification and prediction of pitting corrosion was investigated using one-dimensional convolutional neural networks (1D CNN). Finite element modeling of the pitting corrosion was performed and the probability distribution of the corrosion pits was considered. In the experimental setup, corrosion pits were generated on the corrosion sensor using mechanical drilling. The 1D CNN was adopted to explore the regression relationship between the EMI signatures of the sensor and the mass loss induced by pitting corrosion. The results show that the proposed method has achieved high accuracy in the quantitative prediction of pitting corrosion. This paper lays the technical foundation for real-time and quantitative monitoring of pitting corrosion for metallic structures.

A CNN-integrated percussion method for detection of FRP–concrete interfacial damage with FEM reconstruction

Abstract: Reinforced concrete (RC) structures are commonly strengthened using externally bonded fiber-reinforced polymer (FRP) sheets. The bond between the FRP and concrete is a crucial factor affecting the strengthening effect, and debonding along the FRP–concrete interface is usually accompanied by the fracture of the underlying concrete. Therefore, it is necessary to identify the interface damage of FRP-to-concrete joints and conduct mechanical analysis. However, debonding is invisible damage that occurs inside the underlying FRP layer, which makes damage detection more difficult. To this end, this study fuses a percussion method with a deep learning framework to address the detection of such invisible lesions. Meanwhile, the visualization study provides guidance for later maintenance work. To further illustrate the hazard of the identified lesions, three-dimensional reconstruction for finite element modeling (FEM) with detected damage information based on percussion is proposed to elucidate the mechanical degradation caused by the fracture of underlying concrete. Lastly, the results of this study demonstrate that the detection, visualization, and FEM reconstruction of FRP–concrete interface damage using percussion signals has considerable application potential and is worthy of further study.

Stiffness identification method for asphalt pavement layers and interfaces using monitoring data from built-in sensors

Abstract: The bearing capacity of asphalt pavement gradually deteriorates due to repeated traffic loads. As the crucial mechanical index to evaluate the bearing capacity, the evolution of the stiffness of pavement layers and interfaces is necessary to be mastered. This paper proposes a stiffness identification method for the asphalt pavement layers and interfaces by using monitoring data from built-in sensors. First, the analytical solution of multi-layered elastic/viscoelastic medium with imperfect interlayer subjected to moving load is derived, and the theoretical relationship between stiffness and mechanical response can be obtained. The sensor layout is optimized on the basis of the theoretical relationship. Then, the stiffness identification method is developed, and its feasibility is theoretically explored through an example of three-layered elastic/viscoelastic medium. Finally, the proposed stiffness identification method is applied to a realistic asphalt pavement to validate its reliability. Results show that the proposed method can evaluate the stiffness of pavement layers (including elastic and viscoelastic properties) and interfaces. It is noteworthy that the stiffness identification method by using monitoring data from built-in sensors can be performed in real-time under each passing vehicle load, and is helpful to understand the damage behavior of pavement and guide maintenance decision-making.

Vision-guided unmanned aerial system for rapid multiple-type damage detection and localization

Abstract: Unmanned aerial systems (UASs) are increasingly applied for bridge inspection. A vision-guided UAS with a lightweight convolutional neural network is developed to detect and locate bridge cracks, spalling, and corrosion. The contributions are as follows: (1) To address the problem that traditional UASs are global positioning system (GPS) required while GPS signals under bridge bottom generally are weak. A vision-guided UAS is designed and applied, in which a stereo vision-inertial fusion method is used to provide position data instead of GPS and an ultrasonic ranger is applied to avoid obstacles. (2) Most of the deep learning-based damage detection methods are offline detection, which is unsuitable for UAS-based inspection because the endurance time is limited. To solve this problem, a lightweight end-to-end object detection network is proposed, by replacing the backbone of the original You Only Look Once v3 network with MobileNetv2, and the proposed network of much faster inference speed can be transplanted to the onboard computer of the designed UAS so that real-time edge computing can be performed during inspection. (3) A damage location method based on vision positioning data and simultaneous localization and mapping is also proposed to meet the urgent needs of locating damage in the whole structure. Finally, the proposed system is applied to inspect a long-span bridge to detect and locate the most common damages: crack, spalling, and corrosion with high accuracy and efficiency, which verified the practicability of the system.

An up-scaling temperature compensation framework for guided wave–based structural health monitoring in large composite structures

Abstract: Variations in environmental conditions can significantly impair the accuracy and reliability of guided wave structural health monitoring systems. Acquisition of baseline signals over a wide temperature range for the purposes of damage detection and localization is impractical for large composite structures. A novel framework for compensating the effect of temperature at a post-processing stage is presented in this paper to allow updating the compensation factors using observations obtained at different scales. The proposed methodology utilizes observations collected at the lower scales, where a large amount of data under controlled environment is available. Subsequently, the estimated compensation factors are propagated to the higher scales as priors within a Bayesian framework. This way, the measurements required from the high levels are reduced while making it possible to also update the estimated factors during the operation of the structure. The performance of the methodology is evaluated at different scales and compared with the direct use of compensation factors obtained from coupon studies only. It is demonstrated that the proposed methodology improves the fidelity of the compensation algorithm leading to a reduction in the uncertainty of the temperature-compensated signals. Based on the findings of the present study, the reduction in the uncertainty of the compensation improves the performance of both damage detection as well as damage localization in a large composite panel.

Damage quantification using transfer component analysis combined with Gaussian process regression

Abstract: Machine learning methods used in Structural Health Monitoring applications still have generalization difficulties among structures, even when structures are nominally and topologically similar. The data sets present divergences between their probability distributions that do not allow the model’s generalization for damage detection. This issue is even more complex in situations where one wants to quantify damage levels through data sets collected from different structures. Transfer learning methods offer a solution to overcome those limitations, using relevant information from a labeled structure (source domain) to assist the analysis of another structure (target domain) under unknown conditions. Therefore, this paper proposes the use of transfer component analysis to mitigate divergences between the model/structure’s features, and the label consistency requirement is applied in combination with a Gaussian process regression model for damage quantification. The effectiveness of the estimated model improves when the labels consistency between domains is achieved, indicating the current damage level in the structure when the regression model achieves its best performance (lowest error). The proposed methodology is applied on the benchmark data of a three-story building structure from the Los Alamos National Laboratory using the knowledge from its numerical model under several conditions, where the complete information of its behavior is available. The results compare the analysis in the original space and after applying the proposed methodology, demonstrating an improvement of the performance in the damage detection and quantification steps.

Reweighted-Kurtogram with sub-bands rearranged and ensemble dual-tree complex wavelet packet transform for bearing fault diagnosis

Abstract: The Fast Kurtogram (FK) is a widely used resonance demodulation technique for bearing fault diagnosis. In this paper, a novel method termed Reweighted-Kurtogram with sub-bands rearranged and ensemble dual-tree complex wavelet packet transform (SRE-DTCWPT) is proposed to improve the performance of the FK from the aspects of band division and optimal band selection indicator. To obtain an excellent band division, the SRE-DTCWPT is first developed. It retains the main advantages of DTCWPT and meanwhile addresses the two key issues of frequency sub-bands disorder and frequency bands leakage. Then, a new robust evaluating indicator called reweighted kurtosis is defined. It solves the problem of kurtosis being sensitive to strong impulse interferences. Furthermore, the proposed method involves a set of envelope analysis approaches developed on different cases of fault signals to realize the enhanced identification of the bearing diagnostic information. Two simulated signals and actual bearing signals regarding different practical cases are employed to investigate the effectiveness of the proposed method. In addition, the proposed method is compared with the FK, and the results verify that the proposed method shows high potentials for extracting bearing diagnostic information from complex vibration signals.

Health condition monitoring of bearings based on multifractal spectrum feature with modified empirical mode decomposition-multifractal detrended fluctuation analysis

Abstract: Multifractal detrended fluctuation analysis (MFDFA) is proved to be a powerful tool for fault diagnosis of rotating machinery due to its ability to reveal multifractal structures hidden in nonstationary and nonlinear vibration signals. To overcome the discontinuity of the fitting scale-dependent trend and the poor adaptability of this algorithm, Empirical Mode Decomposition-Multifractal Detrended Fluctuation Analysis (EMD-MFDFA) is introduced. However, EMD-MFDFA runs into difficulties in reverse segmentation and the selection of the expected Intrinsic Mode Functions (IMFs). Aiming at solving these deficiencies, a Modified EMD-MFDFA (MEMD-MFDFA) approach with IMF selection strategy and Step-Moving Window (SMW) segmentation method is proposed in this paper. In MEMD-MFDFA, a metric for distinguishing deterministic and random components is established to select expected IMF components by scaling exponent. Meanwhile, SMW segmentation method is exploited to reduce the estimated errors caused by reverse segmentation. The robustness of the proposed method is investigated through comparing MEMD-MFDFA, MFDFA, and EMD-MFDFA by multifractality of simulated signals with different Signal-to-Noise Ratio (SNR). Furthermore, the proposed approach is applied to three bearing run-to-failure datasets containing three types of faults, and the results show that the multifeatures of the multifractal spectrum obtained by MEMD-MFDFA have the ability to simultaneously identify early fault and assess performance degradation of bearings.

1D-CNN-based damage identification method based on piezoelectric impedance using adjustable inductive shunt circuitry for data enrichment

Abstract: The electromechanical impedance (EMI)-based damage identification method is a non-destructive testing approach in the field of structural health monitoring. The frequency response function (FRF) of EMI can effectively reveal the health conditions of a structure. Typically, the health condition is identified by comparing the FRF of a structure to that of a baseline. However, baselines may exhibit unpredictable shifts in real applications. In this study, a new EMI-based health identification method is proposed without reference to baselines or handcrafted features. An adjustable inductive shunt circuit that can enrich the EMI dataset is connected to a piezoelectric transducer. Pre-set damage, including bolt looseness and mass variations, are selected to demonstrate damage identification. The FRFs are extracted using a phase-sensitive detection algorithm. The damage identification model is realized using a one-dimensional convolutional neural network. Experimental results show that the proposed method can identify the location of bolt loosening and mass variation with an overall accuracy of 99.24%. The proposed method can be applied for identifying the health conditions of a structure with strong nonlinearity.

A novel strategy using optimized MOMED and B-spline based envelope-derivative operator for compound fault detection of the rolling bearing

Abstract: The bearing regularly suffers from compound faults in real-world working conditions. In comparison to the single-fault feature extraction, the compound fault diagnosis is more difficult to achieve. This paper suggests an alternative signal processing strategy using the Multipoint Optimal Minimum Entropy Deconvolution method (MOMED) and B-spline based envelope-derivative operator (EDO) tools. As an upgraded version of the Minimum Entropy Deconvolution tool, the MOMEDA technique has been extensively available for bearing and gear fault detection. However, this approach results in an open problem related to how one can choose an appropriate filter size. Considering this problem, an optimized MOMED based on Salp Swarm Algorithm is proposed. Besides, a novel energy operator method called B-spline based envelope-derivative operator (B-spline EDO) is proposed to detect the corresponding fault characteristics from the two separated mono-component signals produced by the optimized MOMED. The new B-spline EDO method accomplishes higher fault detection performance in a noisy environment. Finally, the experimental results displayed that the novel compound fault detection approach can effectively identify the compound fault characteristics.