This paper studies the problem of bridge health monitoring in an unsupervised manner utilizing only the measured responses from a vehicle passing over a bridge. A half-car model along with a simply ...supported beam is adopted for numerically simulating vehicle–bridge interaction (VBI). Multiple bridge states including healthy bridge and damaged bridge with varying severity at different locations are considered. A data-driven approach based on adversarial autoencoder (AAE) by incorporating the generative capabilities of adversarial autoencoder and pre-processing techniques including frequency filtering and signal averaging is proposed. Vehicle acceleration responses associated only with the healthy bridge state are used for model training. Reconstruction error estimated by the proposed model is adopted as a damage detection index. Along with the detection of the damage in the bridge, the proposed framework is also able to estimate the severity of the damage. The proposed framework also overcomes the limitations of other unsupervised learning approaches such as principal component analysis and autoencoders due to its better representation of data in the latent sub-space with an additional prior distribution constraint. Further, the proposed framework is validated through experimentally obtained data from a laboratory scale bridge model. The contribution of this work is three-fold: First, an adversarial autoencoder-based unsupervised learning framework supplemented by appropriate pre-processing techniques is proposed for drive-by bridge monitoring for the first time, and its implementation is extensively investigated. Second, the superior performance of the proposed AAE framework compared to the competing techniques is demonstrated. Finally, this paper presents one of the early successful attempts of drive-by bridge inspection for monitoring the progressive change in the structure of a bridge. Research presented in this work can potentially open up new opportunities for condition monitoring of bridge networks.
•An adversarial autoencoder-based framework is proposed for drive-by bridge inspection.•The framework is based on unsupervised learning requiring no data from damaged states.•Its superior performance compared to competing techniques is demonstrated.
•Development of a robust algorithm displacement calculation using computer vision algorithms.•Explanation of modifications to action cameras to enable long distance monitoring of ...structures.•Verification of system through intensive laboratory trials with industry leading technology.•Field application of system in uncontrolled conditions with load identification and validation.
This paper describes development of a contactless, low cost vision-based system for displacement measurement of civil structures. Displacement measurements provide a valuable insight into the structural condition and service behaviour of bridges under live loading. Conventional displacement gauges or GPS based systems have limitations in terms of access to the infrastructure and accuracy. The system introduced in this paper provides a low cost durable alternative which is rapidly deployable in the field and does not require direct contact or access to the infrastructure or its vicinity. A commercial action camera was modified to facilitate the use of a telescopic lens and paired with the development of robust displacement identification algorithms based on pattern matching. Performance was evaluated first in a series of controlled laboratory tests and validated against displacement measurements obtained using a fibre optic displacement gauge. The efficiency of the system for field applications was then demonstrated by capturing the validated bridge response of two structures under live loading including the iconic peace bridge. Located in the City of Derry, Northern Ireland, the Peace Bridge is a 310 m curved self-anchored suspension pedestrian bridge structure. The vision-based results of the field experiment were confirmed against displacements calculated from measured accelerations during a dynamic assessment of the structure under crowd loading. In field applications the developed system can achieve a root mean square error (RMSE) of 0.03 mm against verified measurements.
•A data-driven procedure is presented for the Structural Health Monitoring of highway bridges.•The method is based on Gaussian Progress Regression (GPR) and Extreme Function Theory (EFT).•GPR is used ...to estimate the continuous mode shapes from the known values at the sensor locations.•The Extreme Function Theory is then applied for mode shape-based damage detection.•The EFT-based procedure outperforms its EVT-based counterpart in terms of less false positives.
The Extreme Function Theory (EFT) offers a convenient tool for mode shape-based damage detection. When coupled with Gaussian Process Regression (GPR), this statistical framework can provide an automatic and efficient means for Structural Health Monitoring (SHM), especially to reduce the number of false positive errors (i.e. false alarms). Here, the technique is tested experimentally for bridge monitoring purposes on the well-known case study of the I-40 bridge. The EFT-based approach proved able to recognise deviations from the normality model (the undamaged conditions) on this experimental dataset, validating its applicability for large and massive civil structures and infrastructures.
The application of multi-temporal interferometric synthetic aperture radar (MTInSAR) technology in bridge structural health monitoring often encounters considerable challenges due to the intricate ...nature of bridge structures. Notably, the thermal expansion and contraction (TEC) of bridges can lead to prominent interferometric phase jumps at the expansion joints. When the magnitude of the phase jump exceeds π, the continuity assumption required for phase unwrapping is no longer valid. Consequently, classical phase unwrapping methods fail to accurately retrieve bridge deformation. To address this limitation, we propose an adaptive MTInSAR method that can partition the bridge into independent segments and concurrently estimate deformation from multiple reference points. The algorithm first identifies expansion joint locations using a mean square error threshold. Subsequently, reference point selection and segmental phase unwrapping are performed to derive displacement time series of persistent scatterers (PSs), where the mechanical properties of the bridge structure are considered. We validate the effectiveness of the method using 23 TerraSAR-X images of the Shanghai Yangtze River Bridge. The results demonstrate the successful detection of expansion joints and reliable phase unwrapping in PS sub-networks. Moreover, a comparative analysis with the classical minimum cost flow (MCF) method highlights the superior adaptability and reliability of the proposed approach. Finally, threshold values for triggering conditions when phase jumps occur are quantified. The proposed work will enhance the robust monitoring of bridge motions, safeguarding the structural health of bridges.
•First application of low-cost GNSS receivers in bridge deformation monitoring.•Low-cost GNSS results validation against geodetic GNSS and Robotic Total Station.•Low-cost GNSS time series accuracy ...and availability enhancement by Multi-GNSS.•Modal frequencies up to 3 Hz could be identified from low-cost GNSS time-series.•Combined closely-spaced low-cost GNSS solution increases solution robustness.
The development of low-cost GNSS receivers with carrier-phase measurement capacity has led to low-budget GNSS applications of higher accuracy and precision. Recent studies have mainly been carried out with those low-cost receivers for landslide monitoring and achieved promising results. In this study, the performance of two closely-spaced high-rate low-cost GNSS receivers was assessed against the robotic total station (RTS) and geodetic GNSS receiver in monitoring the dynamic response of a major pedestrian suspension bridge at the mid-span. Potential accuracy improvement by the combination of two low-cost GNSS time-series was also examined. It was proved that multi-GNSS solution is required to resolve potential outliers and offsets of the low-cost GNSS time-series, due to cycle slip induced errors. The analysis of the low-cost GNSS time-series showed that the low-cost GNSS receivers can estimate (i) the main dominant frequencies of the bridge with the same accuracy as the geodetic-grade GNSS receiver and (ii) the amplitude of the bridge response with difference of ∼3 mm with respect the geodetic GNSS receiver due to higher noise level. This study revealed the prospect of utilising low-cost GNSS sensors in monitoring dynamic displacement with frequency of 1–3 Hz, corresponding to relatively rigid structures (e.g., short span bridges, etc.).
The multi-temporal satellite-based differential interferometry (MTInSAR) is a well-known remote-sensing technique aimed at detecting displacements of persistent scatterers (PSs) on the terrestrial ...surface. Recent studies motivate research effort on developing efficient and automatic procedures for structural monitoring via MTInSAR. This paper proposes a methodology for the portfolio-scale detection of structural deformations of bridges via MTInSAR. The displacement time-series associated with persistent scatterers on the investigated bridges are used to supply a geoprocessing chain leading to the automatic interpretation of bridge-specific deformation scenarios and the definition of monitoring/assessment priority classes. The methodology is applied to two highway networks in Roma and Bari (Italy) by using Sentinel-1 (C-band) and COSMO-SkyMed (X-band) satellite datasets. Although most of the bridges can be assumed stable, a relevant number of bridges affected by ongoing deformation phenomena are associated with high inspection priority. The PSs and the deformation scenarios related to some specific test bridges subjected to subsidence phenomena are analysed and described in detail. Finally, the deformation scenarios detected through the proposed methodology for two collapsed bridges in Italy are illustrated.
•state-of-the-art on MTInSAR for structural monitoring of bridges is presented.•a methodology (SABRI) for portfolio-scale bridge monitoring via MTInSAR is proposed.•MTInSAR detects several bridges subjected to relevant deformation scenario in Rome.•PS displacement time series on selected test bridges are shown and discussed.•the results of MTInSAR applied on two recently collapsed bridges are described.
•Long-term monitoring data of wind and bridge response were presented from the Hardanger Bridge.•The bridge response exhibited variable results.•The relationship between wind field and bridge ...response was studied using response surface methodology.•Variability in response is attributed to the variable wind field.•The variability in the wind field should be considered in design.
Long-term monitoring data of wind velocities and accelerations on the Hardanger Bridge are used to investigate the relationship between the wind-loading and response processes. The extensive measurement system consisting of 20 accelerometers and 9 anemometers is described as well as the local topography of the site. The wind and response characteristics are presented using scatter plots and wind rose diagrams. The considerable variability observed in the bridge dynamic response is investigated by utilizing response surface methodology. Simple parameters of the wind field are selected as the predictor variables in the analyses. The variability in response is attributed to the variable wind field, and the effects of the significant parameters on the response are presented in a statistical framework. The agreement of the findings with previous considerations and the implications on the design of long-span suspension bridges are discussed.
•An information modeling framework for supporting bridge monitoring applications is proposed.•The framework extends the prior work on the OpenBrIM standards to capture the information relevant to ...engineering analysis and sensor network.•Implementation of the framework employs a NoSQL database system for scalability, flexibility and performance.•The framework is demonstrated using bridge information and sensor data collected from the Telegraph Road Bridge located in Monroe, Michigan.
Bridge management involves a variety of information from different data sources, including geometric model, analysis model, bridge management system (BMS) and structural health monitoring (SHM) system. Current practice of bridge management typically handles these diverse types of data using isolated systems and operates with limited use of the data. Sharing and integration of such information would facilitate meaningful use of the information and improve bridge management, as well as enhance bridge operation and maintenance and public safety. In many industries, information models and interoperability standards have been developed and employed to facilitate information sharing and collaboration. Given the success of building information modeling (BIM) in the Architecture, Engineering and Construction (AEC) industry, efforts have been initiated to develop frameworks and standards for bridge information modeling (BrIM). Current developments of BrIM focus primarily on the physical descriptions of bridge structures, such as geometry and material properties. This paper presents an information modeling framework for supporting bridge monitoring applications. The framework augments and extends the prior work on the OpenBrIM standards to further capture the information relevant to engineering analysis and sensor network. Implementation of the framework employs an open-source NoSQL database system for scalability, flexibility and performance. The framework is demonstrated using bridge information and sensor data collected from the Telegraph Road Bridge located in Monroe, Michigan. The results show that the bridge information modeling framework can potentially facilitate the integration of information involved in bridge monitoring applications, and effectively support and provide services to retrieve and utilize the information.
Visual inspections have been typically used in condition assessment of infrastructure. However, they are based on human judgment and their interpretation of data can differ from acquired results. In ...psychology, this difference is called cognitive bias which directly affects Structural Health Monitoring (SHM)-based decision making. Besides, the confusion between condition state and safety of a bridge is another example of cognitive bias in bridge monitoring. Therefore, integrated computer-based approaches as powerful tools can be significantly applied in SHM systems. This paper explores the relationship between the use of advanced computational intelligence and the development of SHM solutions through conducting an infrastructure monitoring methodology. Artificial Intelligence (AI)-based algorithms, i.e., Artificial Neural Network (ANN), hybrid ANN-based Imperial Competitive Algorithm, and hybrid ANN-based Genetic Algorithm, are developed for damage assessment using a lab-scale composite bridge deck structure. Based on the comparison of the results, the employed evolutionary algorithms could improve the prediction error of the pre-developed network by enhancing the learning procedure of the ANN.
•Forced vibration and in-situ static loading experiments were carried out before service.•Good agreement between the numerical model and the in-situ experiment data.•The warning and critical ...thresholds enables effective judgment for management agencies.•Inclination angles prediction for the 50-year life cycle supports the long-term safety.
In order to reduce the self-weight of Highway No. 4 in the Taichung living circle in Taiwan, a corrugated steel web, with a span of 145m, is used to replace the conventional concrete web. To appraise the structural safety and operating conditions of a prestressed composite box-girder bridge with a corrugated steel web, which is the first bridge of its kind in Taiwan, a bridge monitoring system is developed based on in-situ experiments, numerical modeling, and long-term monitoring. In order to determine the initial static and dynamic behaviors of a real bridge, a series of experiments are first carried out on a newly-constructed bridge. Before entering service, a bridge is subjected to forced vibration experiments and static loading experiments to establish its initial condition. In this study, a numerical model of the bridge is constructed based on the finite element method. The results of the structural analysis are compared with experimental data, and the two sets of results are found to show good agreement. Moreover, thermometers, strain gages, displacement gages, and inclinometers are installed on the bridge to measure changes in the physical quantities, and the monitored temperature gradient profile over a year is fed back to the numerical model for further analysis. Results have indicated that the in-situ linear variable differential transformer (LVDT) and inclinometer monitoring data can be effectively simulated by the numerical model. Finally, based on the material properties, numerical model, and long-term monitoring data from inclinometers, the safety threshold of the bridge is determined to provide a useful reference for bridge management agencies. Prediction of the extreme inclination angles by the Generalized Extreme Value Distribution (GEVD) method for the 50-year life cycle of the monitored bridge also falls within the envelopes of the warning and critical thresholds, which support the long-term safety of bridges.
PDF
