Search

Civil structures may deteriorate during their service life due to degradation or damage imposed by natural hazards such as earthquakes, wind, and impact. Structural performance anomaly detection is essential to provide an early warning of structural degradation limit states in order to prevent potential catastrophic failure. Data-driven machine learning approaches have been widely used for this, due to their capability in capturing features sensitive to damage-induced anomalies from structural health monitoring (SHM) data, assuming that such data are available. Although machine learning models have been used, many are challenged by the vast operational and environmental variability that can corrupt SHM data and by (typically) strongly correlated information from different sensors in the SHM data. This paper proposes an unsupervised deep learning approach for the detection of structural anomaly based on a deep convolutional variational autoencoder (DCVAE) for feature extraction coupled with support vector data description (SVDD) for anomaly detection. The proposed DCVAE-SVDD method has several appealing strengths. First, the variational latent encoding is used to capture the features of monitoring data through a probability distribution. The integration of the Kullback–Leibler divergence in the loss function provides accurate estimation of the probability distributions. Second, the DCVAE designed with convolutional and deconvolutional operations utilizes the correlation among multisensor data to avoid loss of correlation features and achieve better performance in feature extraction. Third, the SVDD is utilized to create a minimum-volume hypersphere that contains the anomaly-sensitive statistical features of the state. The hypersphere accurately separates anomaly-sensitive statistical features of reference states of structure from the anomalous ones. A computational frame model and a laboratory grandstand model are used to evaluate the performance of the proposed method for detecting structural anomaly. The results demonstrate the superiority of the proposed DCVAE-SVDD in detection accuracy over the other commonly used structural anomaly detection methods (deep autoencoder combined with SVDD autoregressive model with one-class support vector machine, and principal component analysis). [ABSTRACT FROM AUTHOR]

Structural health monitoring obtains data reflecting the service status of grid structures through sensors. One of the issues to consider in optimal sensor placement is how to obtain as much information as possible with a limited number of sensors. In this paper, a sensor placement method is proposed based on damage sensitivity and correlation analysis, which is based on strain energy calculation and is suitable for grid structures. Specifically, with the sensor locations as optimization variables, a mathematical optimization model is established by considering the damage sensitivity and redundancy of the monitoring scheme, and a genetic algorithm is employed for computation. Two examples, including a lattice shell and a flat grid, are provided to illustrate the method, followed by a discussion of the sensitivity of parameters such as stiffness reduction degree and load form. The results indicate that the redundancy of the optimized schemes for the two examples decreased by approximately 80% and 30%, respectively. The proposed method ensures a certain degree of damage sensitivity while significantly reducing redundancy, demonstrating its applicability and robustness in sensor placement for grid structures. [ABSTRACT FROM AUTHOR]

Structural anomaly detection based on the structural health monitoring (SHM) data has attracted significant attention owing to its important role in the early warning of structural damage to existing civil structures. Data-driven approaches, where damage-sensitive features are extracted directly from the SHM data using statistical pattern recognition (SPR) techniques without physical models of structures, have been widely studied. Principal component analysis (PCA) and probabilistic PCA (PPCA) are powerful and efficient SPR methods for linear or weakly nonlinear cases. However, some special structures may be subjected to multiple operational conditions, wherein structural configurations such as geometry and mass distribution may change due to the movement of parts or the whole structure, as in retractable roof structures. These changes may give erroneous results in the SPR of the SHM data and eventually in the anomaly detection by a single PCA or PPCA model. This paper presents an improved approach using a mixture of probabilistic principal component analysis (MPPCA) for the anomaly detection of structures under multiple operational conditions with missing measurement data. First, the baseline MPPCA model was constructed for stress data collected under healthy conditions, where the estimation of the MPPCA parameters was reformulated for the missing data cases. Second, three anomaly statistics were presented for newly monitored incomplete data to detect and localize structural anomalies. The probability distributions of the anomaly statistics were estimated to obtain thresholds for outlier detection. Finally, the effectiveness of the MPPCA-based method was investigated by applying the method to the anomaly detection of a retractable roof structure with numerically simulated and real monitored stress data. [ABSTRACT FROM AUTHOR]

Summary: The roof of a large‐span structure is vulnerable to strong winds, and thus, wind‐induced damage of the roof happens occasionally. Structural condition assessment from structural health monitoring (SHM) data provides a good means to analyze the structural safety condition and detect the damage. An accurate state space model is the key point to structural system identification and condition assessment as well. Because the wind loadings on the roof are frequently nonzero and non‐Gaussian, they cannot be directly simplified as random noise for the stochastic subspace identification. Data‐driven combined deterministic‐stochastic subspace identification (CSI) is adopted to identify the state of the wind‐induced vibrating roof structure, in which the wind loadings of the roof are considered as input parameters instead of noise term. Then, a novel specific damage‐sensitive feature is proposed to extract the damage information directly using the system matrices from the identified state space model, which overcomes the diversity at the state matrices. Combined with the statistical pattern recognition (SPR) technique, a statistics‐based damage index is defined to detect the change of the structural state. The proposed method is verified by a numerical simulation of a roof structure, in which the influence of different levels of wind excitation on the methodology is investigated. The results show that the proposed method is sensitive to the change of the structural state but insensitive to wind loading variations. Subsequently, a wind tunnel test performing on a real double‐layer metal roof model further demonstrates the effectiveness of the proposed method for practical application. [ABSTRACT FROM AUTHOR]

Lightweight and flexible membranes offer different façades for buildings (suitability, competitive costs, durability, and other benefits) compared to traditional building materials. Increasing attention is being paid to membrane structures in the civil and industrial sectors. Acquiring response data or environmental characteristics directly from a model or building is the most straightforward approach to analyzing the properties of membrane structures, which also contributes to the development of theoretical studies and simulation methods along with the enactment of specifications. This paper provides a comprehensive overview of membrane structure performance, including mechanical, thermal, and energetic aspects, alongside the deployment and deflation of inflatable types. Furthermore, the devices used to monitor the structural response are summarized. The constitution of the structure is the most critical factor affecting its performance. A proper design would offer enhanced mechanical properties and thermal environments with a reduction in energy consumption. Non-contact measurement technology has the advantage of causing no structural disturbance and is low cost, but it lacks practical application in membrane buildings. The achievements and limitations of previous studies are also discussed. Finally, some potential directions for future work are suggested. [ABSTRACT FROM AUTHOR]

Wireless sensor networks usually suffer from the issue of time synchronization discrepancy due to environmental effects or clock management collapse. This will result in time delays between the dynamic responses collected by wireless sensors. If non-synchronized dynamic response data are directly used for structural modal identification, it leads to the misestimation of modal parameters. To overcome the non-synchronization issue, this study proposes a time synchronization approach to detect and correct asynchronous dynamic responses based on frequency domain decomposition (FDD) with frequency-squeezing processing (FSP). By imposing the expected relationship between modal phase angles extracted from the first-order singular value spectrum, the time lags between different sensors can be estimated, and synchronization can be achieved. The effectiveness of the proposed approach is fully demonstrated by numerical and experimental studies, as well as field measurement of a large-span spatial structure. The results verify that the proposed approach is effective for the time synchronization of wireless accelerometer sensors. [ABSTRACT FROM AUTHOR]

Prestressed structures have gained popularity in large-span buildings due to their great spanning capacity and light self-weights. This kind of structure is normally subjected to multiple types of loads, such as temperature load, wind load, and construction load. The determination of different load effects not only guides the design of similar structures but also helps reveal the damage-induced variation that would be concealed by the environmental loads. To determine the different load effects, separation of the load effects collected by structural health monitoring (SHM) is needed. This study presents an enhanced approach for load effect separation based on independent component analysis (ICA) and ensemble empirical mode decomposition (EEMD), called EEMD-ICA*. The proposed method is to minimize manual tuning of user-defined parameters, which makes the EEMD-ICA suitable for separating load effects from the different measured data. Specifically, an optimization method is developed to determine the appropriate level of the added white noise in the EEMD using the relative root-mean-square error (RMSE) index. A logarithm form of Bayesian information criterion (BIC) is employed for the robust estimation of the number of load effects in the ICA. Simulated structural responses from a square orthogonal cable-net are used to validate the effectiveness of the EEMD-ICA*. Then, the proposed methodology is employed to extract various load effects from the SHM data of the National Speed Skating Oval (NSSO), which is the largest single-layer cable-net structure in the world. [ABSTRACT FROM AUTHOR]

Summary: Structures are subjected to various kinds of structural deterioration and damage with use over long periods of service life. For the safety assurance of structures, it is very important to have a long‐term monitoring system and continuous assessments of structural integrity using the measured data. The objective of this paper is to develop an anomaly detection algorithm for a long‐term structural health monitoring (SHM) system based on probabilistic principal component analysis (PPCA). Static stress data were measured and used in this monitoring system. A baseline PPCA model is built under various environmental loading conditions. Then, newly monitored data are projected onto the principal vectors. Anomaly indices and their probability distributions are evaluated to determine the presence of structural damage indicated by outliers. This method is also capable of dealing with incomplete data and recovering the missing data. First, numerical simulation studies of a revolving auditorium are carried out to validate the proposed PPCA‐based method. Then, real monitoring data collected from the SHM system are used to detect the presence and locations of anomalies in the revolving auditorium. [ABSTRACT FROM AUTHOR]

• A compressive sampling matching pursuit (CoSaMP) algorithm was developed. • The algorithm was used to solve the restoration problem of lossy admittance signatures. • Approach validation was conducted on artificial crack identification of concrete structure. • The approach was compared with iterative shrinkage-thresholding algorithm (ISTA). • Impacts of different loss ratios on restoration accuracy were additionally discussed. • The approach showed effective damage identification using restored admittance signatures. Effective restoration of lossy data is prerequisite to guarantee the reliability of early warning information for damage diagnosis, considering large amounts of sensor data to be collected, transmitted and stored in long-term structural health monitoring (SHM) process. In this study, a compressive sampling matching pursuit (CoSaMP) algorithm was developed for solving the restoration problem of electromechanical admittance (EMA, inverse of impedance) signatures-based damage identification. Observed vectors through a random matrix projection were collected as a substitute of the EMA signatures used in conventional method, and EMA signature restorations from incomplete observed vectors were transformed into solving constrained optimization problems. Then the CoSaMP algorithm was utilized to recover EMA signatures with loss in observed vectors. Validation of the approach was conducted by a series of tests on artificial crack identification for a standardized concrete cube under laboratory conditions, which was compared with that using iterative shrinkage-thresholding algorithm (ISTA)-based algorithm. Additionally, impacts of different loss ratios on restoration accuracy were discussed. The proposed approach showed promising availability for damage identification using EMA technique encountered with lossy data. [ABSTRACT FROM AUTHOR]

In this paper, a customized wireless sensor system (WSS) for structural health monitoring is developed toward large-scale spatial structures. Spatial structures are widely used in large public buildings which generally house thousands of people, therefore the safety of the buildings is a major concern for structural engineers. One of the characteristics of spatial structures is their steel construction and the material is homogeneous throughout. So the strain is commonly the most distinct parameter to indicate status of the structure. Another characteristic of spatial structures is their large-area scale, which brings problem for traditional wired monitoring systems, so an effective wireless sensor network (WSN) for structural monitoring is in urgent demand. Considering those features, the WSS development mainly focused on the sensor selection, hardware design and network customization. In this paper, a vibrating wire sensor is selected for strain measurement because of its stable, durable and anti-electromagnetic properties. For other parameters measurement, some commercial sensor products with digital signal output are adopted. Following the principle of modularization and extendibility, the hardware design is mainly based on the realization of several functional modules. All along, energy efficiency and measurement accuracy are the core design objective. The WSN is classified into four different types of topologies from basic to complex ones. They all have the common working mechanism, namely the collected data transfers via several relay from sensor nodes to sink nodes. Different types of networks are to be customized according to the configuration and scale of different structures. Finally, two typical applications are discussed in detail to verify the feasibility of the system. It can be concluded that the customized WSS is effective and durable, and well satisfies the requirement of structural status monitoring for large-scale spatial structures. Collected data have also shown that the structural stress variation is obvious under the effect of construction process and some other factors. [ABSTRACT FROM AUTHOR]

Advanced structural health monitoring systems have been widely applied to large-span structures for obtaining various structural responses and loads, which are the foundation of performing condition assessment. The structural deformation is continuously employed to estimate the structural condition because it directly provides information about the overall stiffness of the whole structure. Therefore, accurate prediction of structural deformation is essential for reliable assessment of structural conditions. The deformation variation of a large-span prestressed structure is characterized by typical geometric nonlinear effects. This work presents a robust Gaussian process regression (GPR) for building a deformation prediction model for large-span space structures. The proposed approach overcomes the problem of computational cost and employs optimal distribution for modeling noise in monitoring data. Specifically, the PCA method is utilized to reduce the dimension of the input datasets for GPR. The optimal input dataset and noise distribution are estimated via 4 indexes, which are introduced to estimate the prediction performance of deformation prediction models. Simulated structural deformation from Hangzhou Gymnasium is used for verifying the effectiveness of the GPR-based deformation prediction models. Then, the proposed method is employed to predict the vertical deformation of the National Speed Skating Oval (NSSO) during snowfall. Furthermore, the prediction performance of the prediction model is comprehensively investigated via residual analysis. The proposed prediction model could provide a data foundation for the condition assessment of prestressed large-span structures. • Proposed a data-driven prediction model for deformation of large-span prestressed structure based on GPR. • Determined optimal input dataset of prediction model and noise distribution to consider the outlier in monitoring data. • Reduced the computational complexity of GPR-based prediction model via PCA and a moving window strategy. • Revealed the distribution of snow load of the large-span structure by comparing predicted deformations at different points. [ABSTRACT FROM AUTHOR]

Summary: Identification of earthquake ground motion from structural health monitoring (SHM) data provides a good means to reconstruct seismic loads that are essential for postearthquake safety assessments and disaster simulations of structures. Because the data measured by an SHM system are structural absolute response, they cannot be directly applied to the structural motion equation, which is established in relative coordinate system. As such, this paper originally derives the motion equation in absolute coordinate system and then expands the equation into modal space. In addition, the proposed method allows for identifying earthquake ground motion using incomplete modal information and limited measurements through the standard Kalman filter. Subsequently, a numerical two‐dimensional frame is used to validate the feasibility of the proposed method, and the influences of modal parameters and measurement noise on the identification accuracy are also fully investigated. The results show that the proposed method is sensitive to frequency and measurement noise but insensitive to modal shape and damping ratio. It is also found that the identified ground motion subjected to certain measure noise can still be reliably employed for postseismic response calculations of medium‐ and long‐period structures. Finally, a shaking table test performing on a five‐floor frame further demonstrates the effectiveness and accuracy of the proposed identification algorithm for practical application. [ABSTRACT FROM AUTHOR]

• A Lorentz force-based SH-typed EMAT was developed and systematic tested. • A specially design of the flexible circumferential printed circuit was presented. • The designed CPC-EMAT is easy to be assembled and fabricated in real application. • It can generate omnidirectional SH 0 waves with desirable central frequencies. Structural health monitoring (SHM) of in-service structures is becoming increasingly important. The fundamental shear horizontal (SH 0) guided wave mode in plate-like structures shows great potential in damage detection due to its non-dispersive and in-plane vibration properties. In order to generate SH 0 waves, a practical Lorentz force-based electromagnetic acoustic transducer (EMAT) was introduced in this study using the flexible circumferential printed circuit (CPC). The designed principle of CPC-EMAT was similar to that of the circumferential magnet array (CMA)-based EMAT. However, the structure of the CMA-EMAT is complex, and it is difficult to assemble for generating high frequency and uniformly distributed omnidirectional SH 0 waves. Firstly, the performance of the CMA-EMAT with different numbers of magnets was investigated by finite element simulations. Then, the CPC was proposed to replace the CMA with an optimized designed on its size. The CPC-EMAT is easier to fabricate compared to the CMA-EMAT. Finally, experimental tests were conducted for systematic validations on the transducer properties. Simulation and experimental results show that the CPC-EMAT can successfully generate the desirable and acceptable omnidirectional SH 0 waves. The proposed CPC-EMAT is anticipated to find widespread application in SH-typed guided wave-based SHM. [ABSTRACT FROM AUTHOR]

• An improved CMT-BCS method is developed. • The CMT-BCS formulation is restructured to deal with the 'incomplete' CS problem. • The proposed CMT-BCS method has high computational efficiency. • The superiorities of proposed CMT-BCSmethod in accuracy and efficiency are demonstrated. • The influences of the relevant critical parameters are comprehensively explored. The long-term structural health monitoring (SHM) provides massive data, leading to a high demand for data transmission and storage. Compressive sensing (CS) has great potential in alleviating this problem by using less samples to recover the complete signals utilizing the sparsity. Vibration data collected by an SHM system is usually sparse in the frequency domain, and the peaks in their Fourier spectra most often correspond to the same frequencies. This underlying commonality among the signals can be utilized by multi-task learning technique to improve the computational efficiency and accuracy. While being real-valued originally, the data after discrete Fourier transformation are in general complex-valued. In this paper, an improved complex multi-task Bayesian CS (CMT-BCS) method is developed for compression and reconstruction of SHM data requiring a high sampling rate. The novelty of the proposed method is twofold: (i) it overcomes the invalidity of the conventional CMT-BCS approach in dealing with the 'incomplete' CS problem, and (ii) it improves the computational efficiency of conventional CMT-BCS approach. The former is achieved by restructuring the CMT-BCS formulation, and the latter is realized by sharing a common sampling matrix across all tasks of concern. The improved CMT-BCS is evaluated using the shaking table test data of a scale-down frame model and the real-world SHM data acquired from a supertall building. A comparison with several existing BCS methods that enable to deal with complex values is also provided to demonstrate the effectiveness of the proposed method. [ABSTRACT FROM AUTHOR]

• A dictionary is proposed for CS of wind speed using correlation. • A new sliding window is used considering lag, and window size is suggested. • The performance is evaluated using two large-scale spatial structures. • The performance is compared with Fourier basis and linear interpolation. • The influences of relevant critical parameters are comprehensively explored. The real-time wind monitoring is widely used to evaluate the wind effect on the large-scale spatial structures. Wireless sensor network (WSN) is usually the first choice for the large-scale spatial structures to collect wind monitoring data because of its super-large size. Compressive sensing (CS) has great potential in solving the energy problem of WSN and reduces the difficulty in transmission of massive data based on sparsity. However, wind signals are often not naturally sparse on the traditional bases (e.g., Fourier basis). This paper proposes a new method of constructing a dedicated dictionary for wind speed signals using the time-shift strategy. With this proposed dictionary, the signals can be compressed by random sampling and recovered by ℓ 1 -norm sparse regularization. The performance of the improved CS methodology is evaluated using two large-scale spatial structures. The results show that the proposed CS methodology has better performance than the traditional CS algorithm with the Fourier basis and the linear interpolation method. Furthermore, the influences of the relevant critical parameters (regularization parameter, lag, sliding window size, and compression ratio) of the improved CS methodology are comprehensively explored. [ABSTRACT FROM AUTHOR]

• Application of non-convex ℓ p -norm sparse regularization to wind speed data in SHM. • Sparse-sampling method is more robust in sparse signal sampling. • The reconstruction error of this paper is very small. • The wind power density and wind rose before and after compression is consistent. Large-span spatial structures are quite sensitive to wind load because of their notable structural flexibility and low fundamental frequency. Structural health monitoring (SHM) of wind applied to this type of structure is the most direct and effective method of guaranteeing their safety. However, SHM produces a large amount of observation data, and these data often contain compressible redundant information and are usually sparse in the amplitude-frequency domain. To improve their transmission efficiency and quality and explore the characteristics of measured wind load on the surface of a large-span roof, we proposed ℓ p -norm (0 < p < 1) sparse regularization based on compressive sensing for compression and reconstruction of wind speed data in the amplitude-frequency domain. The present compressed data were obtained through a low-rate sparse sampling method according to compressive sensing theory, which is more robust than the traditional sampling method. The alternating direction method of multipliers and the ℓ p shrinkage method were applied to solve nonconvex optimization of reconstructing original data from incomplete measurements. The effectiveness of the proposed method was verified through a field test on a large-span steel roof of a railway station in southern China. The experimental results showed that the proposed method was superior to the smoothed ℓ 0 method and typical ℓ 1 based on the fast iterative shrinkage thresholding method. The reconstruction error was very low; even when the sampling rate was 10%, the signal-to-noise ratio of the reconstruction signal was 21.27, and the absolute error of reconstruction was < 0.05. In addition, the distributions of wind power density and wind rose were consistent before and after compression. [ABSTRACT FROM AUTHOR]