Structural Control & Health Monitoring最新文献

In response to the data anomalies and frequent false alarms caused by harsh environments in long-term structural health monitoring (SHM), this study has reframed the detection of abnormal vibration data as a time series classification problem. This approach identifies multiple anomalous features, thereby reducing manual detection costs. The novel developed Convolutional Neural Network with Squeeze-and-Excitation and Multi-Head Self-Attention (CNN–SE–MHSA) employs a deep residual network structure with channel and spatial attention mechanisms, effectively handling the global long-term dependencies required for anomaly feature learning. It better understands and utilizes feature information across different levels and dimensions, enhancing classification accuracy in complex anomaly situations. Through t-SNE dimensionality reduction visualization and interpretability analysis, it is demonstrated that the model excels in identifying critical features. Furthermore, by generating simulated data with a variational autoencoder (VAE) and implementing transfer learning strategies based on these data, the issue of low recognition accuracy for complex anomaly data due to data imbalance can be effectively mitigated. In a 25-day long-term monitoring experiment of indoor tunnel lining structures, this method demonstrated an average accuracy rate exceeding 96% and a rapid detection capability within 16 min. The results indicate that this method achieves high accuracy in anomaly detection for long-term monitoring data, even when relying exclusively on time-domain data.

Computer vision techniques were employed to monitor the displacement of retaining walls using artificial markers, traditional feature detection algorithms, and photogrammetry-based point cloud reconstruction. However, the use of artificial markers often increases both installation time and costs, whereas the performance of traditional feature matching is affected by uneven illumination, and photogrammetry techniques require multiple images for point cloud reconstruction. To overcome these limitations, a nontarget-based displacement monitoring method for segmental retaining walls (SRWs) using a combination of deep learning and stereovision was proposed. Binocular stereovision was employed to reconstruct the geometry and surface properties of the SRW in a digital three-dimensional (3D) model. Deep learning models were then used to extract natural features from SRW blocks, enabling displacement calculation without using artificial targets. The performance was evaluated by monitoring the behaviors of SRW experiments at both laboratory and field scales. The deep learning–based image segmentation models identified SRW block features in the experiment and real case datasets with an average F1 score from 0.910 to 0.965 under various environmental conditions. The reconstructed results of point cloud coordinates demonstrated high accuracy, ranging from 95.2% to 98.6%. Furthermore, the calculated displacement exhibited a high degree of agreement with the measured displacement. The accuracy of the calculated displacements for the laboratory and field experiments ranged from 89.5% to 99.1%. The proposed method can be used for automatic SRW displacement monitoring.

Monitoring stress and strain at the critical details of steel bridges is essential for ensuring structural integrity. This study introduces a three-layer flexible strain sensor produced through a spraying process, using flake-shaped silver-coated copper powder as the conductive filler and modified acrylic emulsion as the matrix material. The study investigated the impact of size parameters on sensor sensitivity, determining optimal dimensions of 20 mm in length, 2 mm in width, and an initial resistance value ranging from 1.0 Ω to 1.8 Ω. Analysis of the optimized sensor’s performance unveiled high sensitivity and linear response capabilities under low strain conditions with a gauge factor (GF) value of up to 25.6 and a linear correlation coefficient R² ≥ 0.971 under 300 με. Notably, the sensor exhibits an extremely low strain detection limit of 0.005% and a broad response range spanning from 0.005% to 0.19% strain. It demonstrates swift response and recovery times of 500–800 ms, showcases directional strain response, exhibits good repeatability, and endures durability tests (withstanding 3000 cycles). Furthermore, a fitting formula is proposed to accurately depict the strain and relative resistance change relationship across a wide response range. The study and initial application of this sensor’s sensing characteristics and performance signify its potential for practical engineering applications.

Due to their structural characteristics, stay cables are inherently susceptible to vibrations. Addressing this issue, the research explores the dynamics of a two-cable network system, emphasizing the impact of composite vibration control methods. A system consisting of two horizontal cables is presented, each fitted with negative stiffness dampers (NSDs) at their anchored ends and interconnected by a cross-tie. A complex eigenvalue equation, formulated based on displacement boundary conditions and the continuity of displacement and force, is validated through numerical simulations. The multimode damping effects of the dual NSDs and cross-tie on the two-cable network are explored through parameter analysis and optimization. The results demonstrate that reducing the stiffness of the cross-tie improves the fundamental modal damping ratio, whereas increasing its stiffness or positioning it close to the cable’s midpoint enhances the vibration frequency. The incorporation of NSDs into the hybrid system significantly increases the maximum damping ratio while lowering the optimal damping coefficient. This study presents a method for calculating the range of negative stiffness values, providing insights into the selection of installation positions and stiffness for the cross-tie, thereby facilitating the design of highly effective multimode vibration control solutions for stay cables.

Damage identification is a key aspect of structural health monitoring (SHM). However, any measurement of the structural response can be impacted by environmental and operational variations (EOVs), which can affect the system and hinder damage detection. It is therefore important to distinguish between damage-induced changes in structural dynamic properties and changes caused by EOVs. To address this issue, this paper proposes a damage identification method based on nonlinear manifold learning, specifically Laplacian eigenmaps (LEs). The method eliminates the impact of EOVs on the damage index by treating them as embedded variables and does not require the direct measurement of environmental parameters. The Gaussian process regression (GPR) prediction model results in small residuals when the structure is healthy and significant increases when the structure is damaged, demonstrating the effectiveness of the method in removing environmental influences. The proposed method is demonstrated using computer-simulated data, where the environmental conditions have a nonlinear effect on the vibration features. The proposed LE-GPR algorithm is then applied to the Z24 and KW51 bridges and successfully identifies structural damage. The advantage of the proposed approach is its ability to eliminate the effects of ambient temperature and accurately identify structural damage.

This study presents a novel framework for data anomaly detection in stay-cables, aimed at establishing an autonomous monitoring system in cable-stayed bridges. Based on the fact that peaks in the power spectra of cable accelerations appear periodically at constant intervals, we classified the anomalous data into two categories in terms of the data quality and behavioral aspects. The framework provides two thresholds derived from the modal property of stay-cables to identify each anomaly type. To validate the performance of the proposed method, we collected long-term monitoring data from stay-cables in a cable-stayed bridge currently in operation in South Korea. Then, the peak information was extracted by adopting an automatic peak-picking technique. We applied the proposed method to establish thresholds that determine the presence of anomalous data. This study validated that the proposed method can determine anomalous types when new data are used as input.

Seismic isolation technologies for large-span structures have rapidly developed alongside the popularization of the seismic resilience concept. To produce a high-efficiency isolation technology with lower energy dissipation demands, this paper proposes a novel inerter-enabled isolator (IeI) and a tailored input energy reduction-oriented design method. The inerter-based damper within the IeI is developed by combining the dashpot, tuning spring, and two inerters to facilitate the optimization of inerter distribution. Assuming the large-span structure remains linear, the overall seismic input energy of the large-span structure with IeIs and its allocation in the superstructure and additional damping are quantified using stochastic energy analysis. The advantages of the IeI over the conventional linear viscous damper (LVD) isolator are elucidated through dimensionless parametric analysis. Based on the results of parametric analysis, an input energy reduction-oriented design method is proposed for the IeI, along with an easy-to-follow diagram that helps with preliminary design in practical applications. The effectiveness of the IeI and the proposed design method is validated through a design case study of a benchmark large-span structure. The results demonstrate that the IeI reduces the seismic response of large-span structures by simultaneously employing the input energy reduction effect of grounded inerters with the damping-enhancing effect of inerter-based dampers. The proposed design method effectively balances the performance of controlling the large-span structure and the isolator displacement. Under consistent control performance and isolator displacement constraints, the IeI requires much less damping coefficient and energy dissipation capacity than the conventional LVD isolator. Moreover, leveraging the damping enhancement and input energy reduction effects, the IeI achieves comparable control performance to the conventional LVD isolator, even under stricter isolator displacement constraints.

Cracks in civil structures are important signs of structural degradation and may indicate the inception of catastrophic failure. However, most of studies that have employed deep learning models for automatic crack detection are limited to high computational demand and require a large amount of labeled data. Long training time is not friendly to model update, and large amount of training data is usually unavailable in real applications. To bridge this gap, the innovation of this study lies in developing a hybrid method that comprises transfer learning (TL) and low-rank dictionary learning (LRDL) for fast crack detection on concrete surfaces. Benefiting from the availability of preextracted features in TL and a limited number of parameters in LRDL, the training time can be significantly minimized without GPU acceleration. Experimental results showed that the time for training a dictionary only takes 25.33 s. Moreover, this new hybrid method reduces the demand for labeled data during training. It achieved an accuracy of 99.68% with only 20% labeled data. Three large-scale images captured under varying conditions (e.g., uneven lighting conditions and very thin cracks) were further used to assess the crack detection performance. These advantages help to implement the proposed TL-LRDL method on resource-limited computers, such as battery-powered UAVs, UGVs, and scarce processing capability of AR headsets.

Tension force is a crucial indicator in reflecting the stressing state of old cables during the cable replacement process. Even though the vibration-based method is popular in the cable force identification due to its simple calculation process and low cost, the frequency is hard to be recognized with both high time and frequency resolutions attributed to the Heisenberg uncertainty principle, which hinders its application in identifying time-varying cable force. In this paper, a novel quasi-real-time cable force identification method is presented based on a quasi-ideal time-frequency analysis method called multi-synchrosqueezing transform (MSST), by which the cable frequencies can be identified with appreciable time-frequency resolution. To achieve the identification in a real-time manner, an Automatic Frequency Order Identification (AFOI) algorithm is developed to recognize the frequency order automatically depending on the MSST result, in which the interference of fake modes and omitted modes to the identification of the actual frequency order is eliminated to a large extent. The performance of the proposed AFOI algorithm and the quasi-real-time cable force identification method is evaluated on a practical cable replacement engineering case. Results show that the correct orders of the multiple frequencies received from MSST can be identified along the time domain, which demonstrates the effectiveness of the proposed method. The variation of the tension force of not only the replaced cable but also its neighbor cables is estimated with desired time-frequency resolution, which promotes the safety state assessment of a cable in a real-time manner during the replacement process.

The inversion of a high-dimensional stiffness matrix with unknown parameters is time-consuming. In this study, a stiffness separation method is used to solve the large-scale matrix inversion problem. Substructures are isolated from the overall structure by mapping the substructure-related matrix, and the solvable equilibrium equations for the substructures can be established. This method divides the entire stiffness matrix into the sub-stiffness matrices, and the size of the matrix is reduced, thus greatly reducing the stiffness matrix inversion workload. Meanwhile, this paper refines the formulation of the stiffness separation method and presents the compatibility of forces and displacements with the stiffness matrix. A space-truss structure with different damage cases is studied to validate the effectiveness of the proposed method. The division of the structure into single and multi-region scenarios is considered, and the effect of the size and number of substructures on the damage identification is analyzed. These results demonstrate that the stiffness separation method can reduce the computational effort required for analyzing large-scale truss structures.