Figure - available from: Journal of Civil Structural Health Monitoring
This content is subject to copyright. Terms and conditions apply.
Source publication
Structural health monitoring research traditionally focuses on detecting damage in members excluding the possibility of weakened joint conditions. Efficient model-based joint damage detection algorithms demand computationally expensive model that may affect the promptness of detection. Deep learning techniques have recently come up as an efficient...
Similar publications
![Figure 0.1 Hierarchy of AI, ML and DL algorithms (adapted from [13])](profile/Ramanuj-Goswami/publication/348236914/figure/fig2/AS:976664636968961@1609866403995/Hierarchy-of-AI-ML-and-DL-algorithms-adapted-from-13_Q320.jpg)
![Figure 0.3 First 5 bending modes, as considered by Mehrjoo et al.[17]](profile/Ramanuj-Goswami/publication/348236914/figure/fig4/AS:976664641171466@1609866404177/First-5-bending-modes-as-considered-by-Mehrjoo-et-al17_Q320.jpg)
![Figure 0.1 Procedure of model updating as adapted from [25]](profile/Ramanuj-Goswami/publication/348236914/figure/fig5/AS:976664641155073@1609866404288/Procedure-of-model-updating-as-adapted-from-25_Q320.jpg)
![Figure 0.3 4 support damage introduction in the beam adapted from [25]](profile/Ramanuj-Goswami/publication/348236914/figure/fig6/AS:976664641155074@1609866404315/4-support-damage-introduction-in-the-beam-adapted-from-25_Q320.jpg)
![Figure 0.4 Elemental change in stiffness, white zone indicating damage[25]](profile/Ramanuj-Goswami/publication/348236914/figure/fig7/AS:976664641150977@1609866404342/Elemental-change-in-stiffness-white-zone-indicating-damage25_Q320.jpg)
Traditional methods of damage detection on civil structures included approaches depended on human judgement and hence required highly skilled persons. Thus to overcome this problem of damage detection, Machine Learning (ML) algorithms were introduced which were more diligent than the traditional methods and Deep Learning algorithms are usually a su...
Citations
... Similarly, a scalogram image-based health monitoring technique at the joint of steel frames was presented by Avci et al. (2020), Pal et al. (2022), Paral et al. (2020), Sharma and Sen (2020). In the study, the classification of undamaged and various damaged processes was achieved by the CNN algorithm. ...
Joint damage initiates a consequential form of damage in the beam-to-column connection in a steel frame structure. Many traditional damage detection techniques are not suited for such cases. However, available vibration-based methods are unable to provide a general joint damage detection technique that can be applied to all types of structures. The primary objective of this study is to develop a connection damage identification technique for a 3D frame using a convolutional neural network (CNN) model. For that purpose, a five-story steel 3D frame is considered. An impact hammer is utilized to excite the structure and collect acceleration responses at various points under both undamaged and damaged conditions. From these responses, scalogram images are generated, which serve as input for the CNN-based deep learning technique. The results are then compared with those obtained using the AlexNet model. The training and testing results demonstrate that the technique can effectively differentiate between undamaged and damaged classes, showcasing its potential as an automated tool for the health monitoring of frame connections. The robustness of the technique is further computationally verified through environmental variability, along with the localization and severity of the damage.
... CNN have proven successful in evaluating structural conditions through image processing, as demonstrated by several researchers (Modarres et al., 2018;Cha et al., 2017;Sharma and Sen, 2020;Tong et al., 2017). In their work, Cha et al. (2017) effectively utilized CNNs to distinguish concrete cracks through images. ...
... Sampling images for such structures is not only intricate but, at times, impractical or even impossible. In light of this challenge, an alternative approach that utilizes arrays of vibration response signals disguised as 1D/2D/3D images has been proposed to leverage the exceptional classification potential of CNNs (Sharma and Sen, 2020). Inspired by the success of CNNs in identifying damage signatures from noisy structural responses, the current study adopts this approach while circumventing the associated challenges of selecting appropriate damage-sensitive features, as elaborated later. ...
... Taking cues from the human visual cortex, CNNs were originally devised for 2D and 3D data like images and videos, primarily targeting visual recognition tasks. However, their utility has expanded significantly to encompass diverse domains such as speech recognition (Abdel-Hamid et al., 2012), natural language processing (Kim, 2014), classification of electrocardiogram (ECG) beats, power engine fault detection (Ince et al., 2016) and structural damage detection (Abdeljaber et al., 2018;Avci et al., 2017;Sharma and Sen, 2020), for example. ...
This research presents a novel approach proposed for the monitoring of mooring systems in Floating Offshore Wind Turbines (FOWTs), employing a combination of Convolutional Neural Networks (CNNs) and Auto-Regressive (AR) models. CNN finds broad application in monitoring intricate structures, as they adeptly handle noisy response data without necessitating profound domain expertise. The precision of CNNs relies on the extraction of meaningful features from input data, necessitating meticulous data curation and labeling for optimal computational efficiency and accurate estimation. Emphasis is placed on the preference for feature-rich small datasets over voluminous yet sparse datasets, aiming to enable CNNs to discern crucial patterns more effectively and mitigate issues such as overfitting and extensive preprocessing. The novelty of the proposed approach lies in the integration of AR models, which serve to compress data and enhance damage-sensitive characteristics in the input for CNNs. This integration involves deploying regression models fitted to historical responses, parameterized with AR coefficients sensitive to damage, and further classifying severity using CNNs. The sequential nature of this approach addresses challenges such as vanishing/exploding gradients, particularly for extended historical data, while also attenuating the impact of noise and irrelevant information through data compression. The study explores the effectiveness of the coupled AR-CNN method in monitoring FOWT mooring lines, with a specific focus on two levels of damage identification: detection with classification and damage severity across diverse damage and operational scenarios. The modified methodology exhibits superior outcomes by conducting a performance analysis against traditional CNNs and other machine-learning methods, highlighting the potential of the AR-CNN strategy to improve the precision of FOWT mooring line condition monitoring. These findings underscore the AR-CNN strategy's potential to enhance the accuracy of FOWT mooring line condition monitoring.
... Avci et al. (2018) proposed a decentralized 1D-CNNbased VDA applied to wireless sensor networks, improving its real-time performance compared with that of a previous study (Abdeljaber et al., 2017). Recently, Sharma and Sen (2020) conducted a two-dimensional, three-story, three-bay shear frame model joint damage experiment, and multiple types of damage were detected. To overcome the limited data availability, particularly for rare or severe damage cases, Almutairi et al. (2021) utilized the finite element method (FEM) to generate the acceleration data of the undamaged and damaged states. ...
Deep learning approaches have emerged as promising solutions for vibration-based damage assessment. Although these approaches have shown great potential, further investigations are required to apply them to real-world problems, as most studies have been limited to training with experimental and numerical simulation data. To address this, this study examines the feasibility of employing a one-dimensional convolutional neural network (1D-CNN) for damage assessment by utilizing both simulated data and field applications on an actual truss bridge. Extensive parametric studies were conducted to investigate the performance of the model under various architectural configurations, sensor quantities, sensor locations, and degrees of damage. The results of the hyperparameter optimization show that a moderate number of optimizable parameters is essential for the universal applicability of optimized hyperparameters across different situations or configurations. Comparative studies with other machine learning and deep learning algorithms have validated the superior performance of the 1D-CNN in vibration-based damage detection. Finally, the field application demonstrated the robust potential of the 1D-CNN for real-world scenarios, achieving an impressive F1-score of 90.58% even with single-channel measurements.
... Fig. 3. Schematic diagram of the one-dimensional convolutional neural network with an input layer, three convolutional layers with pooling in between, two dense layers with one dropout layer in between and an output layer. We adapt the figure from [34]. ...
Uncertainties are significant in the early vibroacoustic development, e.g., of a car body, to prevent costly modifications close to the start of production. First, engineers must know which uncertain parameters are sensitive: Our previous work identified 170 uncertain parameters being sensitive out of a complex finite element model with 1,300 uncertain parameters – a parameter reduction of approximately 87%. Second, engineers aim to find reliable distributions of these sensitive input parameters for finite element simulations. Finding these distributions is very demanding in a large-scale vibroacoustic model with several connecting parameters, as research already acknowledges regarding simplified connections. In this paper, we address this challenge with neural networks. For this, we use data in the frequency domain. Due to the curse of dimensionality, it is difficult to determine the parameter set of 170 parameters with a neural network. Therefore, we examine the influence of the number of parameters on the performance of neural networks. Furthermore, we train a fully connected feed-forward neural network and compare this to a one-dimensional convolutional neural network. The latter exhibits a better performance. Finally, we show how to determine distributions of the analyzed parameters based on artificial measurement data. Due to this process, we can significantly improve our finite element simulations and show how to deal with the challenge of determining uncertain parameters in a large-scale vibroacoustic finite element model based on data in the frequency domain.
... Sony et al. [22] designed a 1D-CNN to identify multiclass damage using bridge vibration data. 1D CNN was also utilised to detect the change of local structural stiffness and mass based on acceleration from a single sensor [23,24]. ...
Intelligent fault diagnosis (IFD) is essential for preventative maintenance (PM) in Industry 4.0. Data-driven approaches have been widely accepted for IFD in smart manufacturing, and various deep learning (DL) models have been developed for different datasets and scenarios. However, an automatic and unified DL framework for developing IFD applications is still required. Hence, this work proposes an efficient framework integrating popular convolutional neural networks (CNNs) for IFD based on time-series data by leveraging automated machine learning (AutoML) and image-like data fusion. After normalisation, uniaxial or triaxial signals are reconstructed into -channel pseudo-images to satisfy the input requirements for CNNs and achieve data-level fusion simultaneously. Then, the model training, hyperparameter optimisation, and evaluation can be taken automatically based on AutoML. Finally, the selected model can be deployed on a cloud server or an edge device (via tiny machine learning). The proposed framework and method were validated via two case studies, demonstrating the framework’s availability for the automatic development of IFD applications and the effectiveness of the proposed data-level fusion method.
... This paper tests the proposed technique on the Qatar University Grandstand Simulator (QUGS), a new large-scale SHM benchmark problem. This benchmark is considered challenging as detecting joint damages demands computationally expensive models, which may lead to a slower approach [22]. For validation, both noise-free and noisy signals are employed. ...
In classic machine learning-based damage detection algorithms, extracting damage-sensitive features from time series is a challenging issue. Also, this paradigm can delay processing procedures and requires preprocessing. Many efforts have been made to overcome this limitation by expanding deep learning (DL) in structural health monitoring (SHM). However, because most of these systems require considerable measurements during the training step, they are unsuitable for real-time applications. To solve the challenges above, we offer a robust approach using two-dimensional convolutional neural networks (CNNs) and support vector machines (SVMs), merging feature extraction and a rapid classifier at the same time. The method employs a shallow CNN network that receives raw acceleration signals. Both noisy and noise-free datasets are used to verify the hybrid CNN-SVM approach. The results showed an increase in robustness, speed efficiency, and accuracy over traditional machine learning approaches. The results proved efficient, making the algorithm reliable even under high noise conditions.
... For this example, the damage was introduced in the form of semi-rigid connection stiffness reduction. Such type of damage is commonly encountered in real steel structures due to issues such as bolt loosening, improper installation, corrosion, etc (Sharma and Sen, 2020). Therefore, it is an active topic in the structural health monitoring community and has been investigated by various authors (Hou et al., 2021a). ...
Vibration-based building health monitoring is a promising and feasible approach to assess the operational state of building structures in a remote, automated, and continuous fashion; however, efficiently handling high-dimensional vibration signals from multiple sensors and effectively coping with missing/noisy data represent two main technical challenges. In order to overcome these issues, this study proposes a novel, reliable and robust framework, abbreviated CLG-BHM, based on a hybrid deep learning architecture. First, the framework uses a 1D convolutional neural network layer to learn low-dimensional representation vectors of long sensor signals, which preserve underlying structures’ dynamic characteristics. Second, temporal relationships within data are distilled via a Long-Short Term Memory layer. Third, the representation vectors of sensors are aggregated with those of their neighbors in a principled way via a graph attention network layer, resulting in a new latent representation rich in both temporal and spatial information. Finally, the latter is gone through a fully-connected layer to provide damage detection results. The performance and viability of the present method are evidenced via various examples involving a simple lumped mass structure, a semi-rigid steel frame, and an experimental 4-story structure from the literature. Moreover, a robustness study is performed, showing that the method can provide reasonable results with the presence of noisy and missing data.
... For instance, Abdeljaber et al. [50] proposed a novel 1D-CNN with an inherent adaptive design to detect damage of the grandstand simulator using acceleration signals. Sharma et al. [51] introduced a 1D-CNN to process acceleration data to locate weakened joints in a semi-rigid framework. Akshay et al. [52] applies 1D-CNN to Lamb wave-based metal plate damage diagnosis. ...
Cavity defects are usually invisible in concrete, failure to detect and repair them in time could significantly weaken the load-bearing capacity of engineering structures, potentially threatening the structural safety during the construction and operation periods. Therefore, it is essential to explore efficient approaches to detect concrete cavity defects. In this study, we propose a novel method to detect concrete cavity defects based on percussion and deep learning (DL) techniques. First, the percussion sound dataset was collected from concrete specimens with different types of precast cavity defects. Next, an innovative DL framework named the multi-scale convolutional bidirectional memory network (MCBMNet) was developed for classifying these percussion sound signals associated with specified concrete cavity defects. The MCBMNet utilizes the multi-scale convolutional operation to generate various scale features directly from raw percussion sounds, and then employs the bidi-rectional memory units to capture intrinsic feature connections and enhance feature separability, thus improving prediction accuracy and robustness. Finally, two experimental cases, including water-free and water-filled conditions, were designed to test the feasibility of the proposed method considering the influences of humidity environment in reality. Experimental results well demonstrate the excellent performances of the proposed MCBMNet for concrete cavity defect detection, with its average accuracy over 98% in both cases. Additionally, the anti-noise capacity and adaptability of MCBMNet were thoroughly investigated, which outperforms conventional learning algorithms.
... Recent research has investigated a one-dimensional convolutional neural network (1D-CNN) for SHM. Accordingly, the 1D-CNN only requires simple array operations using vibration signals directly; therefore, it can reduce the number of hidden layers (Kiranyaz et al., 2021;Maier et al., 2019;Ni et al., 2020;Sharma & Sen, 2020;Zhang et al., 2019). ...
Structural health monitoring (SHM) has been a continuous interest in civil engineering for decades because of its importance to the smooth operation of structures. Machine learning (ML) and deep learning (DL) methods have been successfully applied for damage detection of structures. However, the traditional ML algorithms heavily depend on the features’ choice and the classifier; therefore, their application and accuracy are often limited for complex structural data. This study aims to develop a novel DL model, namely the one-dimensional convolutional neural network (1D-CNN), for the damage detection of structures using time series data. The 1D-CNN model can automatically extract the features of one-dimensional time series data without manual feature extraction. The proposed 1D-CNN model’s performance is evaluated using two experimental benchmark datasets of large-scale steel frames and numerical datasets of a suspension bridge subjected to seismic excitation via the confusion matrix, accuracy, sensitivity (recall), specificity, precision, F1-score, and area under curve (AUC) under receiver operating characteristic (ROC) curve. The results show that the proposed 1D-CNN model achieves superior accuracy for damage severity and damage localization identifications of structures.
... Image-based DL techniques at the joint of frame structures are also found to be available in the domain (Naresh et al., 2022;Pal et al., 2022;Paral et al., 2020;Sharma & Sen, 2020). In the study, the scalogram images were generated from the frequency domain acceleration data by utilizing the continuous wavelet transform, and the classification of undamaged and different damaged cases was achieved by a convolutional neural network algorithm. ...
The main aim of this study is to present a connection damage identification technique in a plane frame structure using statistical features of vibration data and a support vector machine (SVM)-based ML algorithm. For that purpose, a small-scale laboratory-based single-story plane frame is considered. The damage was incorporated into the structure by making a groove at the connection and the base of the frame was excited, and the acceleration responses were collected from various points. From the responses, the standard deviation, median, mean absolute deviation, root mean square, kurtosis, skewness, approximate entropy, Shannon entropy, and Renyi’s entropy were extracted and utilized as input for the SVM algorithm. The training and testing results depict that the technique can differentiate between undamaged and various damaged classes. It indicates its efficacy as an automation tool for the health monitoring of connections in plane frame structures and could be verified for a large-scale structure.
