Bridge Damage Detection Research Articles

Smart and Automated Infrastructure Management: A Deep Learning Approach for Crack Detection in Bridge Images

Artificial Intelligence (AI) and allied disruptive technologies have revolutionized the scientific world. However, civil engineering, in general, and infrastructure management, in particular, are lagging behind the technology adoption curves. Crack identification and assessment are important indicators to assess and evaluate the structural health of critical city infrastructures such as bridges. Historically, such critical infrastructure has been monitored through manual visual inspection. This process is costly, time-consuming, and prone to errors as it relies on the inspector’s knowledge and the gadgets’ precision. To save time and cost, automatic crack and damage detection in bridges and similar infrastructure is required to ensure its efficacy and reliability. However, an automated and reliable system does not exist, particularly in developing countries, presenting a gap targeted in this study. Accordingly, we proposed a two-phased deep learning-based framework for smart infrastructure management to assess the conditions of bridges in developing countries. In the first part of the study, we detected cracks in bridges using the dataset from Pakistan and the online-accessible SDNET2018 dataset. You only look once version 5 (YOLOv5) has been used to locate and classify cracks in the dataset images. To determine the main indicators (precision, recall, and mAP (0.5)), we applied each of the YOLOv5 s, m, and l models to the dataset using a ratio of 7:2:1 for training, validation, and testing, respectively. The mAP (Mean average precision) values of all the models were compared to evaluate their performance. The results show mAP values for the test set of the YOLOv5 s, m, and l as 97.8%, 99.3%, and 99.1%, respectively, indicating the superior performance of the YOLOv5 m model compared to the two counterparts. In the second portion of the study, segmentation of the crack is carried out using the U-Net model to acquire their exact pixels. Using the segmentation mask allocated to the attribute extractor, the pixel’s width, height, and area are measured and visualized on scatter plots and Boxplots to segregate different cracks. Furthermore, the segmentation part validated the output of the proposed YOLOv5 models. This study not only located and classified the cracks based on their severity level, but also segmented the crack pixels and measured their width, height, and area per pixel under different lighting conditions. It is one of the few studies targeting low-cost health assessment and damage detection in bridges of developing countries that otherwise struggle with regular maintenance and rehabilitation of such critical infrastructure. The proposed model can be used by local infrastructure monitoring and rehabilitation authorities for regular condition and health assessment of the bridges and similar infrastructure to move towards a smarter and automated damage assessment system.

Mode Shape Identification and Damage Detection of Bridge by Movable Sensory System

Indirect method to identify dynamic properties of bridge by a passing vehicle is attractive recently since it does not require pre-installed sensors on bridge. However, the extracted dynamic properties are not accurate enough to identify local damages due to road surface roughness. To overcome this issue, a method to identify mode shapes and detect local damages in a simply supported bridge by movable sensory system is proposed in this study. Instead of using passing vehicles, two stationary vehicles are adopted in this study to perform as movable sensory system. They stay stationary at two different locations on the bridge subject to ambient environmental excitation, and their vertical accelerations are collected to calculate the vertical displacement of the two contact points between vehicles and bridge. Then they move to another two locations and the same procedures are repeated. After all locations along the bridge are tested, a matrix similar to frequency response matrix can be constructed and SVD method is used to extract the mode shapes. The element stiffness is thereafter evaluated by the fundamental mode shape, which can be then used to detect local stiffness reduction in the bridge. Numerical study has been conducted to validate the proposed method, and the effect of bridge damping, vehicle damping, measurement noise, and traffic flow is investigated. Moreover, field testing was conducted on Li-Zi-Wan Bridge and the results showed that the proposed method can re-construct the mode shapes and evaluate element bending stiffness as accurately as conventional direct approach. The proposed method is convenient and easy to implement in practice; and it is more robust because it is not affected by road surface roughness.

Reinforced concrete bridge damage detection using arithmetic optimization algorithm with deep feature fusion

<abstract> <p>Inspection of Reinforced Concrete (RC) bridges is critical in order to ensure its safety and conduct essential maintenance works. Earlier defect detection is vital to maintain the stability of the concrete bridges. The current bridge maintenance protocols rely mainly upon manual visual inspection, which is subjective, unreliable and labour-intensive one. On the contrary, computer vision technique, based on deep learning methods, is regarded as the latest technique for structural damage detection due to its end-to-end training without the need for feature engineering. The classification process assists the authorities and engineers in understanding the safety level of the bridge, thus making informed decisions regarding rehabilitation or replacement, and prioritising the repair and maintenance efforts. In this background, the current study develops an RC Bridge Damage Detection using an Arithmetic Optimization Algorithm with a Deep Feature Fusion (RCBDD-AOADFF) method. The purpose of the proposed RCBDD-AOADFF technique is to identify and classify different kinds of defects in RC bridges. In the presented RCBDD-AOADFF technique, the feature fusion process is performed using the Darknet-19 and Nasnet-Mobile models. For damage classification process, the attention-based Long Short-Term Memory (ALSTM) model is used. To enhance the classification results of the ALSTM model, the AOA is applied for the hyperparameter selection process. The performance of the RCBDD-AOADFF method was validated using the RC bridge damage dataset. The extensive analysis outcomes revealed the potentials of the RCBDD-AOADFF technique on RC bridge damage detection process.</p> </abstract>

Automatic wavelet-based clustering approach for damage detection on railway bridges

This paper addresses unsupervised damage detection in railway bridges by presenting a novel AI-based SHM methodology using traffic-induced dynamic responses. To achieve this goal a hybrid combination of wavelets, PCA and cluster analysis is implemented. Damage-sensitive features from train-induced dynamic responses are extracted and allow taking advantage not only of the repeatability of the loading, but also, of its large magnitude, thus enhancing sensitivity to small-magnitude structural changes. The effectiveness of the proposed methodology is validated in a long-span bowstring-arch railway bridge with a permanent structural monitoring system installed. The methodology proved highly sensitive in detecting early damage, even in case of small stiffness reductions that do not impair structural safety, as well as highly robust to false detections.

An Integrated Approach for Structural Health Monitoring and Damage Detection of Bridges: An Experimental Assessment

The issue of monitoring the structural condition of bridges is becoming a top priority worldwide. As is well known, any infrastructure undergoes a progressive deterioration of its structural conditions due to aging by normal service loads and environmental conditions. At the same time, it may suffer serious damages or collapse due to natural phenomena such as earthquakes or strong winds. For this reason, it is essential to rely on efficient and widespread monitoring techniques applied throughout the entire road network. This paper aims to introduce an integrated procedure for structural and material monitoring. With regard to structural monitoring, an innovative approach for monitoring based on Vehicle by Bridge Interaction (VBI) will be proposed. Furthermore, with regard to material monitoring, to evaluate concrete degradation, a non-invasive method based on the continuous monitoring of the pH, as well as chloride and sulfate ions concentration in the concrete, is presented.

Real-time drive-by bridge damage detection using deep auto-encoder

Structural health condition monitoring of bridge structures has been a concern in the last decades due to their aging and deterioration, in which the core task is damage detection. Recently, the drive-by method has gained much attention as it only needs several sensors installed on the passing vehicle. In this paper, we proposed an automatic damage detection method, which can be exploited in real time when the vehicle is passing the bridge. There are three steps in the proposed method: (1) The vehicle’s framed short-time vibrations instead of full-length data are utilized for training a deep auto-encoder model; at this stage, not commonly used time-domain accelerations of the passing vehicle, but its selected frequency-domain responses are employed to circumvent the influence of noises, (2) For the bridge with unknown health conditions, damage indicators can be extracted from its passing vehicle’s short-time vibration data using the trained model, and (3) The bridge’s health states are determined by real-time extracted damage indicators. To verify the proposed idea, a U-shaped continuous beam and a model truck are used to simulate the vehicle bridge interaction system in engineering. Results showed that the proposed method could identify the bridge’s damage with an accuracy of 86.2% when different severity was considered. In addition, it was observed that higher damage severity could not be revealed by greater values of damage indicators in the laboratory test. Instead, a novel index called identified damage ratios was employed as a reference for assessing the severity of the bridge’s damage. It was shown that with the increase in damage severity, the index would increase and gradually approach 100%.

Frequency Identification and Damage Detection of Bridges Using Virtual Contact-Point Responses of Railway Vehicle

This paper proposed an indirect bridge frequency identification and damage detection method based on the virtual contact-point (CP) responses of railway vehicles moving over the bridge. The virtual CP response is acquired to reduce the inverse effects of the track irregularity and vehicle dynamic components on the indirect bridge monitoring of the vehicle–bridge interaction (VBI) system. By combining the virtual CP responses of multiple connected vehicles, the first three bridge frequencies can be identified when vehicles run at high speed. The synchrosqueezing transform (SST) is then adopted to extract bridge-related dynamic components from the virtual CP response for bridge modal identification. The time-frequency (TF) representation of the virtual CP response using SST is further analyzed for bridge damage detection. The spectral entropy (SE) of the TF representation is introduced to quantify TF changes and serves as a damage index for bridge damage localization. Numerical studies are conducted to investigate the effects of measurement noise, damage conditions, and track irregularity on the effectiveness of the proposed method. The results demonstrate that SE is a good indicator for damage detection, and that the proposed method combined with wavelet denoising has good anti-noise performance and can well locate bridge damage using virtual CP response.

The Use of a Movable Vehicle in a Stationary Condition for Indirect Bridge Damage Detection Using Baseline-Free Methodology

The use of an instrumented scanning vehicle has become the center of focus for bridge health monitoring (BHM) due to its cost efficiency, mobility, and practicality. However, indirect BHM still faces challenges such as the effects of road roughness on vehicle response, which can be avoided when the vehicle is in a stationary condition. This paper proposes a baseline-free method to detect bridge damage using a stationary vehicle. The proposed method is implemented in three steps. First, the contact-point response (CPR) of the stationary vehicle is computed. Secondly, the CPR is decomposed into intrinsic mode functions (IMFs) using the variational mode decomposition (VMD) method. Finally, instantaneous amplitude (IA) of a high frequency IMF is computed. The peak represents the existence and location of the damage. A finite element model of a bridge with damage is created. The results show that the method can identify the damage location under different circumstances, such as a vehicle with and without damping, different speeds of the moving vehicle, different sizes of damage, and multiple damage. A higher speed was found to provide better visibility of damages. In addition, smaller damage was less visible than wider damage.

STRUCTURAL HEALTH MONITORING METHODS OF BRIDGESTRUCTURES: AREVIEWOFEXISTING METHODS WITH RECOMMENDATIONS

Bridges are a crucial part of the transport infrastructure and represent an important asset of a society. One of the crucial parts of that asset goes to monitoring and maintenance since the bridge damage could yield to severe consequences in human lives and material damage. Periodic inspection keeps these structures operational and increases bridge integrity. Different methods have been proposed by different researchers for bridge damage detection based on different theories and methodologies. Various review papers on bridge damage detection are published in the literature in the last few decades. In this paper, a concise review of literature on damage detection in bridge structures based on various methodologies is carried out. The purpose of this paper is to critically examine various methods that have been proposed for bridge damage identification along with some recommendations to improve these methods. The gaps, limitations, and superiority of the existing methods have been identified, and future applications which are important for bridge inspection are covered.

Nonlinear vibration analysis of beam-like bridges with multiple breathing cracks under moving vehicle load

The beam-like bridges with small and medium spans are designed to mainly sustain the moving vehicle. Hence, the dynamic behavior of bridges under vehicular loading can provide important information for the bridge design, assessment, and maintenance. Cracks, as one of the most common damage in the bridge structures, alter the bridge dynamic characteristics and therefore responses to the moving vehicle. This paper proposes a novel semi-analytical approach for nonlinear vibration analysis of the beam-like bridges with breathing cracks subjected to vehicular loading, taking into account the dynamic interactions between the vehicle and cracked bridge. The Dirac’s delta function is utilized to represent the local stiffness change at the crack location, leading to closed-form solutions of mode shapes of the cracked beam. The breathing process of the crack, causing eigenproperties of the beam to change over time, is characterized by the time-varying curvature of the bridge. Subsequently, the nonlinear dynamic equation of the vehicle-cracked bridge interaction system is established, which is solved through a formulated double-iterative numerical scheme. Eigenvalues and dynamic responses of the bridge with open crack presented in the literature are compared with those obtained from the proposed approach for purposes of validation. The effects of various crack parameters, vehicle speed, and bridge surface irregularity are investigated through a parametric study. In addition, the detection of breathing cracks is also discussed. The results show that the proposed approach can be used to effectively obtain the nonlinear vibration characteristics and dynamic responses of the beam-like bridge with general boundary conditions and different crack numbers, locations, depths, and initial status under moving vehicle. Moreover, the approach has the potential to be incorporated into the vibration-based bridge damage detection.

A novel transformer model for surface damage detection and cognition of concrete bridges

Bridges play an important role in modern transportation systems and road networks, and hence it is essential to use various models based on visual inspection to detect and prevent the damages on the surface of bridge structure. However, due to the limitation of traditional models or lack of modelling data, bridge damages are often difficult to be accurately detected. This paper proposed a novel deep learning model called Bridge Detection Transformers (BR-DETR) based on Detection Transformers (DETR). Through analysis of existing bridge damage instances, we used a copy-paste data augmentation method to create new samples and significantly increased the sample size. Convolution was replaced by Deformable Conv2D, which introduces two-dimensional offsets to the regular grid sampling positions of standard convolution. Convolutional Project Attention was also added after the self-attention layer, which enabled additional modeling of local spatial context. In each encoder and decoder layer, Locally-enhanced Feed-Forward (LeFF) was used to replace the Feedforward to promote the correlation between adjacent tokens in the spatial dimension. The BR-DETR model outperformed the DETR model in detection performance with increased mAP and recall on the augmented highway bridge damage dataset and on the augmented Shandong bridge damage dataset.

Damage detection of in-service steel railway bridges using a fine k-nearest neighbor machine learning classifier

Minor areas of surface corrosion in steel railway bridges can grow progressively andlead to localized section losses and structural failure over time. This paper proposes a novel combined damage detection approach for the classification of various extents and degrees of cross section losses due to damages like corrosion using a k-Nearest Neighbor (kNN) machine learning classifier. A Finite Element (FE) model of an in-service railway bridge is developed and validated using vibration data from field testing and these combined FE-field data are trained and tested to classify various corrosion cases following the Australian Standard AS7636. The results show that the proposed technique is practical and highly accurate in classifying damages in steel railway bridges, even if a minor level of steel corrosion is intended to be classified. Furthermore, a comparison between the accuracies of kNN classifier and the Radial Basis Function (RBF) Gaussian kernel Support Vector Machine (SVM) is presented.

Application of couple sparse coding in smart damage detection of truss bridges

Damage detection in bridge structures plays a crucial role in the ‘in-time’ maintenance of such structures, which subsequently prevents further propagation of the damage, and likely collapse of the structure. Currently, the applications of machine learning algorithms are growing in the smart damage detection of structures. This work focuses on application of a new machine learning algorithm to identify the location and severity of damage in truss bridges. Frequency response functions are used as damage features, and are compressed using principal component analysis. Couple sparse coding is adopted as a classification method to learn the relationship between the bridge damage features and its damage states. Two truss bridges are used to test the proposed method and determine its accuracy in the detection of damage to truss bridges. It is found that the proposed method provides a reliable detection of damage location and severity in truss bridges.

Railway bridge damage detection based on extraction of instantaneous frequency by Wavelet Synchrosqueezed Transform

In bridge structural health monitoring, typically the dynamic response of the system is used to assess the health condition of the bridge. However, the dynamic interaction between a bridge and a passing vehicle imposes non-stationarity on the system response, whereby the bridge modal parameters become time-dependent and detecting damage, for example, based on the bridge modal parameters, becomes challenging. Dynamic vehicle-bridge interaction (VBI) responses have mainly been investigated for damage detection through identifying signal singularities and abrupt changes. The singularities are usually associated with high-frequency components (relative to the bridge natural frequencies), and it is demanding to isolate the damage-induced singularities from those caused by either an operational condition, i.e., track irregularities, or noise. Unlike the high-frequency range, the influence of damage on the resonance frequency of the coupled system has not been fully explored. The present study proposes the shape of the bridge instantaneous frequency as a damage sensitive feature in which the influence of the vehicle dynamics can be excluded. This study demonstrates the feasibility of a damage detection approach based on the bridge instantaneous frequency by applying Wavelet Synchrosqueezed Transform (WSST). In this approach the bridge instantaneous frequency variation induced by damage is distinguished from the bridge instantaneous frequency variation induced by the vehicle. Several damage scenarios that are implemented numerically are analyzed to verify the method’s performance. The results demonstrate that a high resolution instantaneous frequency extracted from the VBI dynamic response outperforms the resonance frequency in determining the local disruption, leading to detecting the damage. A Damage Index (DI) is also proposed as an attempt to quantify the damage severity.

Structural damage detection of old ADA steel truss bridge using vibration data

This study proposes a statistical-based detection method that is responsive to damage and not to environmental and operational conditions. The method serves as a damage recognition system for structural health monitoring using field measurements from real bridges. Vehicle-induced bridge and ambient vibration measurements collected from the benchmark Old ADA steel truss bridge situated in Japan were utilized to validate the proposed method. The steel truss members in the bridge were subjected to five different damage scenarios to represent common potential problems in structural health monitoring of real-life applications. The collected measurements have been completely published and made available online. A combination of principal component analysis (PCA) and linear discriminant analysis (LDA) transformation is utilized as a statistical-based recognition technique. Vibration data representing power spectral density (PSD) functions were tested as damage-sensitive features from identified condition sources. The proposed combination of the PCA-LDA transformation system outperforms the popular PCA transformation as a statistical model for classification of state conditions. Although the first two principal components of PCA hold 50–85% of the variation in data, the first two components from PCA- LDA hold about 95% of the total variation. As a result, the three PCs, of PCA-LDA, visualization successfully managed to classify the five structural damage scenarios into their five individual subgroups.

Damage Detection of Steel Truss Bridges Based on Gaussian Bayesian Networks

This paper proposes the use of Gaussian Bayesian networks (GBNs) for damage detection of steel truss bridges by using the strain monitoring data. Based on the proposed damage detection procedure, a three-layer GBN model is first constructed based on the load factors, structural deflections, and the stress measurements of steel truss bridges. More specifically, the load factors of the structures are defined as the first-layer network nodes, structural deflections are considered as the second-layer network nodes, and the third-layer nodes of the GBN model are built based on the stress data of the truss elements. To achieve the training for the constructed GBN model, the finite element analysis of the bridge structures under the different load factors is performed. Then, the training of the network is performing by using the maximum likelihood estimation approach, and the optimized network parameters are obtained. Based on the trained network model, the measured load factors and the corresponding stress monitoring data of a limited number of truss elements are considered as input, and the stress measurements of all truss elements of bridges can be accurately estimated by searching the optimized topological information among network nodes. For a steel truss bridge, when the truss elements are damaged, the stress states of the damaged elements will be changed. Therefore, a damage index is further constructed for damage detection of steel truss bridges based on the changed stress states of those damaged elements. To verify the feasible and effective use of the proposed damage detection approach, an 80 m steel truss bridge with various damage cases was conducted as numerical simulations, and the investigation results show that the trained GBN can be accurately used for stress prediction of steel truss bridges, and the proposed damage index with the estimated stress data can be further applied for structural damage localization and quantification with a better accuracy. Furthermore, the results also suggest that the proposed damage detection procedure is accurate and reliable for steel truss bridges under vehicle loads.

Review of artificial intelligence-based bridge damage detection

Bridges are often located in harsh environments and are thus extremely susceptible to damage. If the initial damage is not detected in time, it can develop further causing safety hazards. Therefore, accurate detection of bridge damage is an important topic. In recent years, artificial intelligence technology has been developed rapidly, especially machine learning algorithms, which have shown amazing results in various fields while it also received attention in bridge inspection. This paper summarizes the progress of bridge damage detection research related to artificial intelligence techniques between 2015 and 2021. For structural health monitoring, sensing data is the basis for various data processing methods. The strength and weakness of the sensing data itself directly affect the effectiveness of subsequent processing methods. As a result, this paper classifies bridge damage detection studies into six categories from the types of sensing data: visual image, point cloud, infrared thermal imaging, ground-penetrating radar, vibration response, and other types of data. These six types of damage detection methods were reviewed and summarized respectively. Finally, challenges and future trends were discussed.

Unmanned Aircraft System Applications in Damage Detection and Service Life Prediction for Bridges: A Review

The increasing need for inexpensive, safe, highly efficient, and time-saving damage detection technology, combined with emerging technologies, has made damage detection by unmanned aircraft systems (UAS) an active research area. In the past, numerous sensors have been developed for damage detection, but these sensors have only recently been integrated with UAS. UAS damage detection specifically concerns data collection, path planning, multi-sensor fusion, system integration, damage quantification, and data processing in building a prediction model to predict the remaining service life. This review provides an overview of crucial scientific advances that marked the development of UAS inspection: underlying UAS platforms, peripherals, sensing equipment, data processing approaches, and service life prediction models. Example equipment includes a visual camera, a multispectral sensor, a hyperspectral sensor, a thermal infrared sensor, and light detection and ranging (LiDAR). This review also includes highlights of the remaining scientific challenges and development trends, including the critical need for self-navigated control, autonomic damage detection, and deterioration model building. Finally, we conclude with a brief discussion regarding the pros and cons of this emerging technology, along with a prospect of UAS technology research for damage detection.

Bridge damage detection using operating deflection shape ratios obtained from a passing vehicle

Large scale monitoring of bridge structures poses a significant challenge for infrastructure managers. This challenge could be overcome by using passing vehicles to monitor bridge condition. The operating deflection shapes of a bridge can theoretically be detected from in-vehicle measurements. This paper proposes a novel approach which introduces the concept of operating deflection shape ratios (ODSRs) from the responses measured on two following axles on a bridge. Unlike operating deflection shapes derived from drive-by measurements, ODSRs are non-dimensional values and are not a function of the bridge excitation level. This means that ODSRs are always constant for various vehicles passages. It is theoretically demonstrated in this study that ODSRs corresponding to each mode of bridge vibration can be approximately extracted from the responses of two following axles. Finite element modelling is employed to validate the feasibility of using ODSRs for bridge monitoring. The sensitivity of the ODSRs to damage is demonstrated numerically for damage at various locations, different vehicle speeds and different pavement roughness values. Laboratory experiments are then used to verify the theory in a practical environment. It is confirmed that the ODSR can be calculated from in-vehicle vibration measurements and can be used to detect changes in the structural behaviour of the bridge. The results of the study indicate that the ODSR represents a useful feature for drive-by bridge monitoring.

Application Research of Bridge Damage Detection Based on the Improved Lightweight Convolutional Neural Network Model

To ensure the safety and rational use of bridge traffic lines, the existing bridge structural damage detection models are not perfect for feature extraction and have difficulty meeting the practicability of detection equipment. Based on the YOLO (You Only Look Once) algorithm, this paper proposes a lightweight target detection algorithm with enhanced feature extraction of bridge structural damage. The BIFPN (Bidirectional Feature Pyramid Network) network structure is used for multi-scale feature fusion, which enhances the ability to extract damage features of bridge structures, and uses EFL (Equalized Focal Loss) to optimize the sample imbalance processing mechanism, which improves the accuracy of bridge structure damage target detection. The evaluation test of the model has been carried out in the constructed BDD (Bridge Damage Dataset) dataset. Compared with the YOLOv3-tiny, YOLOv5S, and B-YOLOv5S models, the mAP@.5 of the BE-YOLOv5S model increased by 45.1%, 2%, and 1.6% respectively. The analysis and comparison of the experimental results prove that the BE-YOLOv5S network model proposed in this paper has a better performance and a more reliable performance in the detection of bridge structural damage. It can meet the needs of bridge structure damage detection engineering with high requirements for real-time and flexibility.