Vibration-Based Structural Damage Detection with a Convolutional Transformer Network

Vibration-based damage detection methods, leveraging structural response data to identify structural damage, play an important role in structural health monitoring, particularly for critical infrastructure like bridges and high-rise buildings. In recent years, deep learning, which can autonomously learn damage-sensitive features and classify damage from complex datasets, has been widely used in structural damage detection. This paper reviews deep learning architectures, including Convolutional Neural Networks (CNNs), Recurrent Neural Networks (RNNs), and autoencoders, applied in vibration-based structural damage detection for civil structures over the past ten years, highlighting their advantages and disadvantages. The challenges and future trends for vibration-based damage detection by using deep learning are summarized.

In the past few decades, vibration-based structural damage detection (SDD) has attracted widespread attention. Using the response data of engineering structures, the researchers have developed many methods for damage localization and quantification. Adopting meta-heuristic algorithms, in which particle swarm optimization (PSO) is the most widely used, is a popular approach. Various PSO variants have also been proposed for improving its performance in SDD, and they are generally based on the Global topology. However, in addition to the Global topology, other topologies are also developed in the related literature to enhance the performance of the PSO algorithm. The effects of PSO topologies depend significantly on the studied problems. Therefore, in this article, we conduct a performance investigation of eight PSO topologies in SDD. The success rate and mean iterations that are obtained from the numerical simulations are considered as the evaluation indexes. Furthermore, the average rank and Bonferroni-Dunn’s test are further utilized to perform the statistic analysis. From these analysis results, the Four Clusters are shown to be the more favorable PSO topologies in SDD.

Truncation error analysis on modal flexibility-based deflections: application to mass regular and irregular structures

The effect of structural uncertainties or measurement errors on damage detection results makes the robustness become one of the most important features during identification. Due to the wide use of vibration signatures on damage detection, the development of vibration-based techniques has attracted a great interest. In this work, a review on vibration-based robust detection techniques is presented, in which the robustness is considerably improved through modeling error compensation, environmental variation reduction, denoising, or proper sensing system design. It is hoped that this study can give help on structural health monitoring or damage mitigation control.

In vibration-based damage detection in structures, often changes in the dynamic properties such as natural frequencies, modeshapes, and derivatives of modeshapes are used to identify the damaged elements. If only a partial list of natural frequencies is known, optimization methods may need to be used to identify the damage. In this research, the algorithm proposed by Podlevskyi & Yaroshko (2013) is used to determine the stiffness distribution in shear building models. The lateral load resisting elements are presented as a single equivalent spring, and masses are lumped at floor levels. The proposed method calculates stiffness values directly, i.e., without optimization, from the known partial list of natural frequency data and mass distribution. It is shown that if the number of stories with reduced stiffness is smaller than the number of known natural frequencies, the stories with reduced stiffnesses can be identified. Numerical studies on building models with two stories and four stories are used to illustrate the solution method. Effect of error or noise in given natural frequencies on stiffness estimates and, conversely, sensitivity of natural frequencies to changes in stiffness are studied using 7-, 15-, 30-, and 50-story numerical models. From the studies, it is learnt that as the number of stories increases, the natural frequencies become less sensitive to stiffness changes. Additionally, eight laboratory experiments were conducted on a five-story aluminum structural model. Ten slender columns were used in each story of the specimen. Damage was simulated by removing columns in one, two, or three stories. The method can locate and quantify the damage in cases presented in the experimental studies. It is also applied to a 1/3 scaled 18-story steel moment frame building tested on an earthquake simulator (Suita et al., 2015) to identify the reduction in the stiffness due to fractures of beam flanges. Only the first two natural frequencies are used to determine the reductions in the stiffness since the third mode of the tower is torsional and no reasonable planar spring-mass model can be developed to present all of the translational modes. The method produced possible cases of the softening when the damage was assumed to occur at a single story.

Structural health monitoring (SHM) based on long-term safety and efficacy has attracted much attention. The success of SHM methods depends on the information content of the measurements; in addition, there are challenges in processing SHM in terms of the optimal number and location of sensors. In this study, we provide a convenient method for structural damage detection using vibration data from plane steel frames. We apply 1-D convolutional neural networks (1D-CNNs) to present an influential sensor selection analysis to maintain performance and optimize the number and location of sensors. Finding the damage location based on the optimal number and location of sensors will significantly reduce the detection time and the number of installed sensors. The proposed structural damage detection method uses stress analysis to find positions with large displacements, then divides the detection zones, selects highly discriminate sensors, analyzes the influence of sensor combinations within each zone to achieve optimization of sensor number as well as position selection, and establishes CNN models of the overall structure for damage detection of the joints. The proposed method was practically validated on benchmark data from the Qatar University Grandstand Simulator (QUGS). The experimental results show that the sensor demand rate is only 16.67% compared to the previous optimal number and location of sensors with no restrictions on the mounting position. Moreover, only four CNN models are required to predict damage/undamaged for the entire structure (30 connections). The accuracy of damage detection is improved to 96.62% using sensor combinations with influence.

Few-shot learning (FSL) is of great significance to the field of machine learning. The ability to learn and generalize using a small number of samples is an obvious distinction between artificial intelligence and humans. In the FSL domain, most graph neural networks (GNNs) focus on transferring labeled sample information to an unlabeled query sample, ignoring the important role of semantic information during the classification process. Our proposed method embeds semantic information of classes into a GNN, creating a word embedding distribution propagation graph network (WPGN) for FSL. We merge the attention mechanism with our backbone network, use the Mahalanobis distance to calculate the similarity of classes, select the Funnel ReLU (FReLU) function as the activation function of the Transform layer, and update the point graph and word embedding distribution graph. In extensive experiments on FSL benchmarks, compared with the baseline model, the accuracy of the WPGN on the 5-way-1/2/5 shot tasks increased by 9.03, 4.56, and 4.15%, respectively.

Few-shot learning is a method to acquire learning ability in a small amount of sample data scenarios. This paper aims to study an online and offline interaction model applied to international Chinese education and teaching based on the few-shot learning method. It first expounds an overview of international Chinese education, including the connotation and characteristics of the hybrid mode of online teaching and offline teaching in international Chinese education, then designs an online and offline interactive model, and finally compares it with a baseline model and a traditional single teaching model. The experimental data shows that the accuracy and recall rate of the international Chinese education online and offline interaction model based on few-shot learning both reach 70%, which verifies its effectiveness.

This study proposes an analytical model for vibrations in a cracked rectangular plate as one of the results from a program of research on vibration based damage detection in aircraft panel structures. This particular work considers an isotropic plate, typically made of aluminum, and containing a crack in the form of a continuous line with its center located at the center of the plate and parallel to one edge of the plate. The plate is subjected to a point load on its surface for three different possible boundary conditions, and one examined in detail. Galerkin’s method is applied to reformulate the governing equation of the cracked plate into time dependent modal coordinates. Nonlinearity is introduced by appropriate formulations introduced by applying Berger’s method. An approximate solution technique—the method of multiple scales—is applied to solve the nonlinear equation of the cracked plate. The results are presented in terms of natural frequency versus crack length and plate thickness, and the nonlinear amplitude response of the plate is calculated for one set of boundary conditions and three different load locations, over a practical range of external excitation frequencies.

Structural damage detection is very important for identifying and diagnosing the nature of the damage in an early stage so as to reduce catastrophic failures and prolong the service life of structures. In this paper, a novel approach is presented that integrates independent component analysis (ICA) and support vector machine (SVM). The procedure involves extracting independent components from measured sensor data through ICA and then using these signals as input data for a SVM classifier. The experiment presented employs the benchmark data from the University of British Columbia to examine the effectiveness of the method. Results showed that the accuracy of damage detection using the proposed method is significantly better than the approach by integrating ICA and ANN. Furthermore, the prediction output can be used to identify different types and levels of structure damages.

ABSTRACTThis study presents damage localisation in steel cantilever beams including multiple cracks based on changes in the modal curvature and flexibility. Natural frequencies and corresponding mode shapes of circular-hollow-sectional and box-sectional steel cantilever beams are obtained by analytical and experimental models under six damage scenarios. Based on these data, the modal curvature and modal flexibility methods are employed to localise the cracks within the beams. Comparisons are made to show effectiveness of the methods in determination of crack locations.

This study addresses the problem of maize disease detection in agricultural production, proposing a high-accuracy detection method based on Attention Generative Adversarial Network (Attention-GAN) and few-shot learning. The method introduces an attention mechanism, enabling the model to focus more on the significant parts of the image, thereby enhancing model performance. Concurrently, data augmentation is performed through Generative Adversarial Network (GAN) to generate more training samples, overcoming the difficulties of few-shot learning. Experimental results demonstrate that this method surpasses other baseline models in accuracy, recall, and mean average precision (mAP), achieving 0.97, 0.92, and 0.95, respectively. These results validate the high accuracy and stability of the method in handling maize disease detection tasks. This research provides a new approach to solving the problem of few samples in practical applications and offers valuable references for subsequent research, contributing to the advancement of agricultural informatization and intelligence.

Two major challenges associated with a vibration-based damage detection method using changes in natural frequencies are addressed: accurate modeling of structures and the development of a robust inverse algorithm to detect damage, which are defined as the forward and inverse problems, respectively. To resolve the forward problem, new physics-based finite element modeling techniques are developed for fillets in thin-walled beams and for bolted joints, so that complex structures can be accurately modeled with a reasonable model size. To resolve the inverse problem, a logistic function transformation is introduced to convert the constrained optimization problem to an unconstrained one, and a robust iterative algorithm using the Levenberg-Marquardt method is developed to accurately detect the locations and extent of damage. The new methodology can ensure global convergence of the iterative algorithm in solving under-determined system equations and deal with damage detection problems with relatively large modeling error and measurement noise. It is applied to various engineering structures including lightning masts, a space frame structure and one of its components, and a pipeline. The exact locations and extent of damage can be detected in the numerical simulation, and the locations and extent of damage can be successfully detected in experimental damage detection.

The most successful machine learning technology considered for analyzing a significant amount of chest X-ray images is Deep Learning and it has the potential to cause significant influence on Covid-19 screening. In this paper, we analyze four distinct Convolutional Neural Network (CNN) state-of-the-art architectures that are Baseline Model, Vanilla CNN, VGG-16 and Siamese Model on the basis of test accuracy. The effectiveness of the models under consideration is assessed using the Chest Radiograph dataset, which is publicly available for research. In order to discover COVID-19, we used well-known deep learning algorithms for data rarity. These include employing Siamese networks using transfer learning and a few-shot learning approach. Our experiments show that using few-shot learning methodologies, we can create a COVID-19 identification model that is both efficient and effective even with limited data. With this strategy, we were able to achieve 95% accuracy, compared to 86% with Baseline model.

Stroke is a major factor in causing disability and fatalities. Doctors use computerized tomography (CT) and magnetic resonance imaging (MRI) scans to assess the severity of a stroke. Automatic image segmentation can help doctors diagnose strokes more quickly and accurately, but it is challenging due to the variability of stroke lesions and the limited availability of labeled data. Deep learning is the cutting-edge technique of machine learning and artificial intelligence, which needs an extensive labeled dataset for effective training. Unfortunately, in the medical domain, the availability of labeled data is severely limited, posing a challenge for conventional deep- learning approaches. In this article, we introduce a system that utilizes deep learning in the form of fusing transformer-based and convolutional neural network (CNN)-based features and few-shot learning techniques to segment ischemic strokes in multimedia MRIs. To accomplish this, we employ two different methods. The first method involves parallel fusion, where we combine CNN-based and transformer-based features. The second method utilizes serial fusion, combining CNN-based and transformer models using few-shot learning. Through the integration of transformer and CNN models, we can extract both global and local features and enhance the system's performance. Moreover, we tackle the issue of limited labeled data by integrating few-shot learning techniques. Additionally, our system optimizes efficiency by selecting only the slices with lesions, disregarding unlesioned slices. The system under consideration is trained with the BraTS2020 dataset, evaluated on the ISLES 2015 dataset, and contrasted the performance with cutting-edge systems. The suggested system attains a dice coefficient score of 0.76, surpassing the scores of previous cutting-edge systems by a substantial margin.