Structural performance assessment considering both observed and latent environmental and operational conditions: A Gaussian process and probability principal component analysis method

Structural faults like damage and degradations will cause changes in structure response data. Performance assessment can be conducted by investigating such changes. In real implementations however, structural responses are affected by environmental and operational variations (EOVs) as well. Such variation should be well captured by the assessment model when detecting structural changes. It should be noted that not all EOVs can be measured by the monitoring system. When both observed and latent EOVs have significant effects on the monitored structural responses, these two effects should be considered properly. Furthermore, uncertainties can be significant for the monitoring data since loads and EOVs cannot be directly controlled under working conditions. To address these problems, this work proposes a performance assessment method considering both observed and latent EOVs. A Gaussian process is used to model the functional behaviour between structural response and observed EOVs whilst principal component analysis is used to eliminate the effect of latent EOVs. These two methods are combined using a Bayesian formulation and the effect of both observed and latent EOVs are modelled. The associated model parameters are inferred through probability density functions to account for the uncertainties. A synthetic data example is presented to validate the proposed method. It is also applied to the monitoring data of a long-span cable-stayed bridge with different damage scenarios considered, illustrating its capability of real implementations in structural health monitoring.

Diagnosis of internal cracks in concrete using vibro-acoustic modulation and machine learning

This paper investigates the utility of physics-informed machine learning models for vibro-acoustic modulation (VAM)–based damage localization in concrete structures. Vibro-acoustic modulation is a nonlinear dynamics-based non-destructive testing method, which was initially developed to perform damage detection and later extended to accomplish damage localization. The VAM-based damage (hidden crack) diagnosis is performed by analyzing the damage index pattern on the surface of the component to arrive at the size and location of the hidden damage. Past investigations have employed heuristically selected damage index thresholds as well as computationally expensive Bayesian estimation methods for VAM-based damage localization in two (surface) dimensions. Compared to these studies, the proposed methodology automates the threshold selection (algorithmic instead of heuristic), increases the speed of the probabilistic damage diagnosis process, and enables the estimation of damage depth. We generate training data (damage index) for the machine learning models using the pertinent nonlinear dynamics (finite element) models using different combinations of test parameters. The (supervised) machine learning models are thus informed by computational physics models. These include two types of artificial neural network (ANN) models: classification models that identify whether a sensor location is damaged or not and regression models that enable Bayesian estimation to obtain the posterior probability distribution of damage location and size. The accuracy of machine learning-based diagnosis is evaluated using both numerical and laboratory experiments. The proposed physics-informed machine learning models for VAM-based damage diagnosis are able to achieve an accuracy of about 60–64% in the validation experiments, indicating the potential of these methods for internal crack detection. The results show that for complex (nonlinear dynamics-driven) diagnostic methods, damage index patterns learned from physics models could be successfully used for damage detection as well as localization.

A two-step computer vision-based framework for bolt loosening detection and its implementation on a smartphone application

Bolt loosening detection is a labor-intensive and time-consuming process for field engineers. This paper develops a two-step computer vision-based framework to quickly identify bolt loosening angle from field images captured by unmanned aerial vehicle (UAV). In step one, a total of 1200 image samples of bolted structures were used to train faster region based convolutional neural network (Faster R-CNN) for bolt detection from UAV captured images. In step two, computer vision-based technologies, including Gaussian filter, perspective transform, and Hough transform (HT), were performed to quantify bolt loosening angle. The developed framework was then integrated into web server and an iOS application (app) was designed to enable fast data communication between field workplace (UAV captured images) and web server (bolt loosening angle quantification), so that field engineers can quickly view the inspection results on their phone screens. The proposed framework and designed smartphone app greatly help field engineers to improve the accuracy and efficiency for onsite inspection and maintenance of bolted structures.

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

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.

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

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.

Damage detection in cementitious materials with optimized absolute electrical resistance tomography

Electrical resistance tomography (ERT) serves as a non-invasive, non-destructive, non-radioactive imaging technique. It has potential applications in industrial and biological imaging. This paper presents an optimized inverse algorithm, named Newton’s Constrained Reconstruction Method (NCRM), to detect damage in cementitious materials. Several constraints were utilized in the proposed algorithm to optimize initial parameters. The range and spatial distribution of conductivities within the sample were chosen as two main constraints. Two sets of numerical and a set of experimental voltage data were used to reconstruct conductivity distribution images based on this algorithm. To evaluate the quality of reconstructed images, two image quality evaluation indicators, correlation coefficient and position error were used. Results show that the proposed algorithm NCRM has the ability to enhance the reconstructed image quality with fewer artifacts and has better positioning accuracy.

Sensor data-based probabilistic monitoring of time-history deflections of railway bridges induced by high-speed trains

Structural condition monitoring of railway bridges has been emphasized for guaranteeing the passenger comfort and safety. Various attempts have been made to monitor structural conditions, but many of them have focused on monitoring dynamic characteristics in frequency domain representation which requires additional data transformation. Occurrence of abnormal structural responses, however, can be intuitively detected by directly monitoring the time-history responses, and it may give information including the time to occur the abnormal responses and the magnitude of the dynamic amplification. Therefore, this study suggests a new Bayesian method for directly monitoring the time-history deflections induced by high-speed trains. To train the monitoring model, the data preprocessing of speed estimation and data synchronization are conducted first for the given training data of the raw time-history deflection; the Bayesian inference is then introduced for the derivation of the probability-based dynamic thresholds for each train type. After constructing the model, the detection of the abnormal deflection data is proceeded. The speed estimation and data synchronization are conducted again for the test data, and the anomaly score and ratio are estimated based on the probabilistic monitoring model. A warning is generated if the anomaly ratio is at an unacceptable level; otherwise, the deflection is considered as a normal condition. A high-speed railway bridge in operation is chosen for the verification of the proposed method, in which a probabilistic monitoring model is constructed from displacement time-histories during train passage. It is shown that the model can specify an anomaly of a train-track-bridge system.

High resolution bolt pre-load looseness monitoring using coda wave interferometry

This paper proposes a new concept, named the Detectable Resolution of Bolt Pre-load (DRBP), by using the coda wave interferometry (CWI) to quantitatively measure the pre-load looseness at a high resolution. Due to its characteristics of roughness, irregularity, and randomly distributed asperities, the contact surface of the bolted components can function as a natural interferometer to scatter the propagation waves. The multiply-scattered coda waves can amplify the slight changes in the travel path and show the visible perturbation in the time domain. By calculating the time-shifted correlation coefficient of coda waves before and after the slight pre-load looseness, the tiny pre-load changes can be clearly revealed. To evaluate the feasibility of the proposed method, a theoretical model considering the time shifts of coda waves and the variations of pre-load is established. Based on the acoustoelastic effect and the wave path summation theory of coda wave interferometry, the model shows that the time shifts of coda waves change linearly with the variations of pre-load. Verification experiments are conducted, and the results show that the R-square values of the fitting curves are larger than 0.9216. In addition, the proposed approach has the feature of high resolution. The ultimate pre-load resolution of the proposed approach is 0.331%, that is, when the variation of pre-load is larger than 0.331%, it can be detected. Therefore, theoretical analysis and experimental results prove that the CWI-based pre-load detection approach holds great potential for the detection of bolt pre-load looseness, especially during the initial stage.

A baseline-free method for damage identification in pipes from local vibration mode pair frequencies

As structural systems approach their end of service life, integrity assessment and condition monitoring during late life becomes necessary in order to identify damage due to age-related issues such as corrosion and fatigue and hence prevent failure. In this paper, a novel method of level 3 damage identification (i.e. detection, localisation and quantification) from local vibration mode pair (LVMP) frequencies is introduced. Detection is achieved by observation of LVMP frequencies within any of the vibration modes investigated while the location of the damage is predicted based on the ranking order of the LVMP frequency ratios and the damage is quantified in terms of material volume loss from pre-established quantification relations. The proposed method which is baseline-free (in the sense that it does not require vibration-based assessment or modal data from the undamaged state of the pipe) and solely frequency-dependent was found to be more than 90% accurate in detecting, locating and quantifying damage through a numerical verification study. It was also successfully assessed using experimental modal data obtained from laboratory tests performed on an aluminium pipe with artificially inflicted corrosion-like damage underscoring a novel concept in vibration-based damage identification for pipes.

Probability of detection, localization, and sizing: The evolution of reliability metrics in Structural Health Monitoring

The successful implementation of Structural Health Monitoring (SHM) systems is confined to the capability of evaluating their performance, reliability, and durability. Although there are many SHM techniques capable of detecting, locating and quantifying damage in several types of structures, their certification process is still limited. Despite the effort of academia and industry in defining methodologies for the performance assessment of such systems in recent years, many challenges remain to be solved. Methodologies used in Non-Destructive Evaluation (NDE) have been taken as a starting point to develop the required metrics for SHM, such as Probability of Detection (POD) curves. However, the transposition of such methodologies to SHM is anything but straightforward because additional factors should be considered. The time dependency of the data, the larger amount of variability sources and the complexity of the structures to be monitored exacerbate/aggravate the existing challenges, suggesting that much work has still to be done in SHM. The article focuses on the current challenges and barriers preventing the development of proper reliability metrics for SHM, analyzing the main differences with respect to POD methodologies for NDE. It was found that the development of POD curves for SHM systems requires a higher level of statistical expertise and their use in the literature is still limited to few studies. Finally, the discussion extends beyond POD curves towards new metrics such as Probability of Localization (POL) and Probability of Sizing (POS) curves, reflecting the diagnosis paradigm of SHM.