The DOI matches its Crossref record; 42 of the 42 references it cites were found in a registry or an index.
Lost data reconstruction for structural health monitoring using deep convolutional generative adversarial networks
Abstract
In the application of structural health monitoring, the measured data might be temporarily or permanently lost due to sensor fault or transmission failure. The measured data with a high data loss ratio undermine its ability for modal identifications and structural condition evaluations. To reconstruct the lost data in the field of structural health monitoring, this study proposes a deep convolutional generative adversarial network which includes a generator with encoder–decoder structure and an adversarial discriminator. The proposed generative adversarial network model needs to understand the content of the complete signals, as well as produce realistic hypotheses for the lost signals. Given the data stably measured before the occurrence of data loss, the generator is trained to extract the features maintained in the data set and reconstruct lost signals using the responses of the remaining functional sensors alone. The discriminator feeds back the distinguished results to the generator to improve its reconstruction accuracy. When training the model, the reconstruction loss and the adversarial loss are employed to better handle the low-frequency features and high-frequency features of the signals. The effectiveness and efficiency of the proposed method are validated by two case studies. As the number of training epoch increases, the reconstructed signals learn the features from low-frequency to high-frequency, and the amplitude of the reconstructed signals gradually increases. It can be seen that the final reconstruction signals match well with the real signals in the time domain and frequency domain. To further demonstrate the applicability of the reconstructed signals in data analysis, the reconstructed acceleration data are used to accurately identify the modal parameters in the numerical case, and the vehicle-induced responses are precisely decomposed from the reconstructed strain data in the field case. Finally, the reconstruction capacity is also investigated with the different numbers of the faulted strain gauges.
Overview
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
Claims other papers make about this one
Each is checked against this paper’s own text: does it say what the citing paper says it does?
In the citing paper
“Lei et al. [79] exploited the DCGAN to learn the features of the pre-loss data to reconstruct the lost data. Then, they distinguished the actual and reconstructed data using discriminators to improve the reconstruction accuracy.”
From Deep Learning for Structural Health Monitoring: Data, Algorithms, Applications, Challenges, and Trends (2023) · cites this paper as [79]
Citations
Cites
Proceedings of IEEE Conference on Computer Vision and Pattern Recognition
(1993) · doi:10.1109/cvpr.1993 · Cited by 17,667
Deep Learning‐Based Crack Damage Detection Using Convolutional Neural Networks
Cha, Y., Choi, W., & Büyüköztürk, O. (2017) · Computer-Aided Civil and Infrastructure Engineering · doi:10.1111/mice.12263 · Cited by 3,295
Cha, Y., Choi, W., Suh, G., Mahmoudkhani, S., & Büyüköztürk, O. (2018) · Computer-Aided Civil and Infrastructure Engineering · doi:10.1111/mice.12334 · Cited by 1,558 · cited 2×
“Cha et al. 25 implemented region-based CNNs to accurately detect and localize concrete cracks and steel corrosion, which have greatly advanced the performance in damage detection.”
BackgroundVibration Based Condition Monitoring: A Review
Carden, E. P., & Fanning, P. (2004) · Structural Health Monitoring · doi:10.1177/1475921704047500 · Cited by 1,543 · cited 1×
“Structural health monitoring (SHM) techniques have been widely applied in bridges by installing structural health monitoring systems (SHMSs) to measure structural responses and evaluate structural conditions.(1,2) The practical application of SHMS consists of various types of sensors, such as accelerometers, strain ga…”
MethodAccelerating Very Deep Convolutional Networks for Classification and Detection
Zhang, X., Zou, J., He, K., & Sun, J. (2016) · IEEE Transactions on Pattern Analysis and Machine Intelligence · doi:10.1109/tpami.2015.2502579 · Cited by 832 · cited 1×
“Figure 4 illustrates four main types of nonlinear activation functions used in the deep learning, that is, leaky rectified linear unit function (LReLU), rectified linear unit function (ReLU), hyperbolic tangent function (Tanh), and sigmoid function (Sigmoid).(39) LReLU and ReLU perform better in alleviating gradient v…”
BackgroundSun, L., Shang, Z., Xia, Y., Bhowmick, S., & Nagarajaiah, S. (2020) · Journal of Structural Engineering · doi:10.1061/(asce)st.1943-541x.0002535 · Cited by 731 · cited 1×
“Using frequency-domain analysis, time–frequency domain analysis, and statistical analysis, structural safety and performance are evaluated from these measured signals.(3) However, in practice, data loss inevitably occurs which seriously affects structural measurements and assessments.”
MethodSelf-citationTang, Z., Chen, Z., Bao, Y., & Li, H. (2019) · Structural Control and Health Monitoring · doi:10.1002/stc.2296 · Cited by 436 · cited 1×
“Tang et al. 24 trained CNNs with time and frequency information of measured signals for data anomaly detection.”
Abdulkarem, M., Samsudin, K., Rokhani, F. Z., & A Rasid, M. F. (2020) · Structural Health Monitoring · doi:10.1177/1475921719854528 · Cited by 424
Atha, D. J., & Jahanshahi, M. R. (2018) · Structural Health Monitoring · doi:10.1177/1475921717737051 · Cited by 372 · cited 1×
“14 In the SHM field, CNNs have been widely employed to address crack detection, 15,16 corrosion detection, 17 vehicle detection, 18 and structural condition evaluation.”
MethodWang, Z., & Cha, Y.-J. (2021) · Structural Health Monitoring · doi:10.1177/1475921720934051 · Cited by 336 · cited 2×
“Cha et al. 25 implemented region-based CNNs to accurately detect and localize concrete cracks and steel corrosion, which have greatly advanced the performance in damage detection.”
MethodCrack and Noncrack Classification from Concrete Surface Images Using Machine Learning
Kim, H., Ahn, E., Shin, M., & Sim, S.-H. (2019) · Structural Health Monitoring · doi:10.1177/1475921718768747 · Cited by 315 · cited 1×
“37 As for the output sizes of convolutional layers, it is calculated by the sizes of the input layer, filter kernel, zero pad-ding, and sliding stride with the following equation where O C denotes the output height/length, W C denotes the input height/length, K C denotes the filter size, P C…”
MethodVibration‐based structural state identification by a 1‐dimensional convolutional neural network
Zhang, Y., Miyamori, Y., Mikami, S., & Saito, T. (2019) · Computer-Aided Civil and Infrastructure Engineering · doi:10.1111/mice.12447 · Cited by 303 · cited 2×
“Zhang et al. 21 trained one-dimensional (1D) CNN models with measured acceleration data to detect local structural stiffness and mass changes and validated the proposed CNNs on two experimental beam models and one actual bridge structure.”
MethodDeep learning for data anomaly detection and data compression of a long‐span suspension bridge
Ni, F., Zhang, J., & Noori, M. N. (2020) · Computer-Aided Civil and Infrastructure Engineering · doi:10.1111/mice.12528 · Cited by 264 · cited 1×
“Ni et al. 28 proposed a novel data compression and reconstruction method using autoen-coders.”
MethodBayesian multi-task learning methodology for reconstruction of structural health monitoring data
Wan, H.-P., & Ni, Y.-Q. (2019) · Structural Health Monitoring · doi:10.1177/1475921718794953 · Cited by 240 · cited 1×
“Wan and Ni 11 introduced the Bayesian multi-task learning approach to reconstruct the lost temperature data and acceleration signals.”
BackgroundDeep learning based on Batch Normalization for P300 signal detection
Liu, M., Wu, W., Gu, Z., Yu, Z., Qi, F., & Li, Y. (2018) · Neurocomputing · doi:10.1016/j.neucom.2017.08.039 · Cited by 230 · cited 1×
“38 It normalizes original batch inputs x i to the zero-mean and unity-variance distribution with their mean values m and variances s 2 .”
BackgroundDeep learning-based automatic volumetric damage quantification using depth camera
Beckman, G. H., Polyzois, D., & Cha, Y.-J. (2019) · Automation in Construction · doi:10.1016/j.autcon.2018.12.006 · Cited by 209 · cited 1×
“14 In the SHM field, CNNs have been widely employed to address crack detection, 15,16 corrosion detection, 17 vehicle detection, 18 and structural condition evaluation.”
MethodLost data recovery for structural health monitoring based on convolutional neural networks
Fan, G., Li, J., & Hao, H. (2019) · Structural Control and Health Monitoring · doi:10.1002/stc.2433 · Cited by 187 · cited 1×
“Fan et al. 27 employed auto-encoders with skip connection to reconstruct the incomplete measured vibration data with a high loss ratio.”
MethodConvolutional neural network–based data recovery method for structural health monitoring
Oh, B. K., Glišić, B., Kim, Y., & Park, H. S. (2020) · Structural Health Monitoring · doi:10.1177/1475921719897571 · Cited by 165 · cited 1×
“Oh et al. 20 proposed a CNN-based data reconstruction method for SHM using a converted grid form of measured data.”
MethodVibration-based damage detection for bridges by deep convolutional denoising autoencoder
Shang, Z., Sun, L., Xia, Y., & We, Z. (2020) · Structural Health Monitoring · doi:10.1177/1475921720942836 · Cited by 164 · cited 1×
“26 This feature is quite suitable for time-series data compression and reconstruction.”
BackgroundSelf-citationDetection, identification, and quantification of sensor fault in a sensor network
Kullaa, J. (2013) · Mechanical Systems and Signal Processing · doi:10.1016/j.ymssp.2013.05.007 · Cited by 163 · cited 1×
“including sensor fault, transmission failure, radio interaction, power failure, and aging of sensor components.(4) A comprehensive wired or wireless SHMS demands quite a few sensors, transmission systems, and data acquisition modules and controllers.”
BackgroundFan, G., Li, J., & Hao, H. (2020) · Structural Health Monitoring · doi:10.1177/1475921720916881 · Cited by 154
Seismic response prediction method for building structures using convolutional neural network
Oh, B. K., Park, Y., & Park, H. S. (2020) · Structural Control and Health Monitoring · doi:10.1002/stc.2519 · Cited by 143 · cited 2×
“Oh et al. 20 proposed a CNN-based data reconstruction method for SHM using a converted grid form of measured data.”
MethodBao, Y., Tang, Z., & Li, H. (2019) · Structural Health Monitoring · doi:10.1177/1475921719844039 · Cited by 140 · cited 1×
“Many algorithms have been proposed to reconstruct the lost data in SHM. Bao et al. 9 proposed the compressive sensing technique to recover and estimate the lost signals in wireless systems.”
Damage detection in a novel deep-learning framework: a robust method for feature extraction
Guo, T., Wu, L., Wang, C., & Xu, Z. (2019) · Structural Health Monitoring · doi:10.1177/1475921719846051 · Cited by 116
Zhou, Y., & Sun, L. (2018) · Structural Health Monitoring · doi:10.1177/1475921718773954 · Cited by 110 · cited 1×
“36 When dynamic loads are applied to the structures, the responses can be measured with installed sensors (accelerometers, strain gauges, inclinometers, etc.).”
MethodSelf-citationRestoring method for missing data of spatial structural stress monitoring based on correlation
Zhang, Z., & Luo, Y. (2017) · Mechanical Systems and Signal Processing · doi:10.1016/j.ymssp.2017.01.018 · Cited by 101 · cited 1×
“Zhang and Luo 13 also used the correlation analysis between structural responses to recover the lost data in building structures.”
Chen, Z., Li, H., & Bao, Y. (2018) · Structural Health Monitoring · doi:10.1177/1475921718788703 · Cited by 91 · cited 2×
“Chen et al.(10) employed the copula method to capture the relationship in structural responses and reconstruct the lost signals.”
MethodSeno, A. H., Khodaei, Z. S., & Aliabadi, M. H. (2019) · Mechanical Systems and Signal Processing · doi:10.1016/j.ymssp.2019.04.023 · Cited by 81 · cited 1×
“34,35 Before training the deep learning models, it is important to use data cleaning techniques to clean and transform the raw data into an understandable format.”
MethodCancelli, A., Laflamme, S., Alipour, A. A., Sritharan, S., & Ubertini, F. (2019) · Structural Health Monitoring · doi:10.1177/1475921718820015 · Cited by 77 · cited 1×
“Stochastic subspace identification (SSI) is a robust output-only modal identification technique to extract dynamic characteristics.(44) It is employed to identify the natural frequencies, damping ratios, and mode shapes of this numerical cable-stayed bridge model from the collected real signals and reconstructed signa…”
BackgroundXia, Y., Jian, X., Bin, Y., & Su, D. (2019) · Remote Sensing · doi:10.3390/rs11222651 · Cited by 68 · cited 1×
“14 In the SHM field, CNNs have been widely employed to address crack detection, 15,16 corrosion detection, 17 vehicle detection, 18 and structural condition evaluation.”
MethodSelf-citationBridge-Deflection Estimation through Inclinometer Data Considering Structural Damages
Zhang, W., Sun, L., & Sun, S. W. (2016) · Journal of Bridge Engineering · doi:10.1061/(asce)be.1943-5592.0000979 · Cited by 66 · cited 2×
“Zhang et al. 21 trained one-dimensional (1D) CNN models with measured acceleration data to detect local structural stiffness and mass changes and validated the proposed CNNs on two experimental beam models and one actual bridge structure.”
MethodSelf-citationA novel distribution regression approach for data loss compensation in structural health monitoring
Chen, Z., Bao, Y., Li, H., & Spencer, B. F. (2017) · Structural Health Monitoring · doi:10.1177/1475921717745719 · Cited by 62 · cited 1×
“The measured signals with completely lost data or data with high loss ratios are difficult to reflect the real structural behaviors.(7) In particular, the neutral axis should be calculated with two parallel strain gauges; if one strain gauge fails, the other cannot be used to evaluate the neutral axis on that section.”
BackgroundCNN-based damage identification method of tied-arch bridge using spatial-spectral information
Duan, Y., Chen, Q., Zhang, H., Yun, C. B., Wu, S., & Zhu, Q. (2019) · Smart Structures and Systems · doi:10.12989/sss.2019.23.5.507 · Cited by 54 · cited 1×
“Duan et al. 23 introduced two-dimensional (2D) CNNs to develop a method detecting the hanger damages of tied-arch bridges.”
MethodBridge condition monitoring under moving loads using two sensor measurements
Nie, Z., Lin, J., Li, J., Hao, H., & Ma, H. (2019) · Structural Health Monitoring · doi:10.1177/1475921719868930 · Cited by 53 · cited 1×
“Structural health monitoring (SHM) techniques have been widely applied in bridges by installing structural health monitoring systems (SHMSs) to measure structural responses and evaluate structural conditions.(1,2) The practical application of SHMS consists of various types of sensors, such as accelerometers, strain ga…”
MethodDeep semi-supervised generative adversarial fault diagnostics of rolling element bearings
Verstraete, D., Droguett, E. L., Meruane, V., Modarres, M., & Ferrada, A. (2019) · Structural Health Monitoring · doi:10.1177/1475921719850576 · Cited by 52 · cited 1×
“30 The generative network generates candidates while the discriminative network evaluates them.”
BackgroundXia, Y., Wang, P., & Sun, L. (2018) · International Journal of Structural Stability and Dynamics · doi:10.1142/s0219455419400157 · Cited by 37 · cited 1×
“8 Although replacing the faulty sensors is a way to solve these problems, it requires a labor-intensive and time-consuming process to discover and address the issues.”
BackgroundSelf-citationBao, Y., Guo, Y., & Li, H. (2020) · Structural Health Monitoring · doi:10.1177/1475921720909440 · Cited by 37 · cited 1×
“The time-domain signals are transferred into the frequency domain using the fast Fourier transform (FFT).(41) Figure 10 shows the comparison of real and reconstructed signals for the A7 accelerometer with the number of training epochs.”
MethodKim, S. G., Chae, Y. H., & Seong, P. H. (2020) · Annals of Nuclear Energy · doi:10.1016/j.anucene.2020.107410 · Cited by 36 · cited 1×
“Kim et al. 32 proposed GANs to reconstruct multiple lost structural signals.”
BackgroundSeno, A. H., & Aliabadi, M. H. (2019) · Sensors · doi:10.3390/s19173659 · Cited by 35 · cited 1×
“34,35 Before training the deep learning models, it is important to use data cleaning techniques to clean and transform the raw data into an understandable format.”
MethodWu, B., Li, H., & Huang, Y. (2018) · Structural Health Monitoring · doi:10.1177/1475921718790212 · Cited by 30 · cited 1×
“33 Deep learning is an effective tool to automatically extract the features and correlations from raw data.”
MethodConstrained observability method in static structural system identification
Lei, J. W., Nogal, M., Lozano-Galant, J. A., Xu, D., & Turmo, J. (2017) · Structural Control and Health Monitoring · doi:10.1002/stc.2040 · Cited by 26 · cited 1×
“42,43 Since the dynamic characteristics are sensitive to structural damages, it is necessary to ensure the modal identification results of the reconstructed signal are very close to the ground truth.”
BackgroundA Generative Adversarial Network Model for Simulating Various Types of Human-Induced Loads
Xiong, J., & Chen, J. (2019) · International Journal of Structural Stability and Dynamics · doi:10.1142/s0219455419500925 · Cited by 26 · cited 2×
“Xiong and Chen 31 employed GANs to generate all types of human-induced loads, including walking, jumping, and bouncing loads.”
Background
Research works printed but not linked: 3
- Khodabandehlou H, 2019, Struct Health Monit, 26, e2308
- Duan YF, 2019, Smart Struct Syst, 23, 507
- Lei J, 2019, Struct Health Monit, 26
Cited by
Deep learning-based structural health monitoring
Cha, Y.-J., Ali, R., Lewis, J., & Büyükӧztürk, O. (2024) · Automation in Construction · doi:10.1016/j.autcon.2024.105328 · Cited by 418
Zhang, J., Huang, M., Wan, N., Deng, Z., He, Z., & Luo, J. (2024) · Measurement · doi:10.1016/j.measurement.2024.114528 · Cited by 160
Jia, J., & Li, Y. (2023) · Sensors · doi:10.3390/s23218824 · Cited by 182
“Specifically, constructing models, collecting data, and detecting damage in the laboratory is called the experimental phase.”
BackgroundTashakori, A., Jiang, Z., Servati, A., Soltanian, S., Narayana, H., Le, K., Nakayama, C., Yang, C.-L., Wang, Z. J., Eng, J. J., et al. (2024) · Nature Machine Intelligence · doi:10.1038/s42256-023-00780-9 · Cited by 137
“These factors, in addition to lack of washing or sanitization methods, lead to limited practical usability and accuracy of these solutions 1,4,42 .”
Sepasdar, R., Karpatne, A., & Shakiba, M. (2022) · Computer Methods in Applied Mechanics and Engineering · doi:10.1016/j.cma.2022.115126 · Cited by 137
Metadata
- Authors
- Lei, X., Sun, L., & Xia, Y.
- Affiliations
- Tongji University (CN)
- Corresponding author
- Xia, Ye
- Venue
- Structural Health Monitoring
- Publisher
- SAGE Publications
- Published
- 2020-10-10
- Volume
- 20
- Issue
- 4
- Pages
- 2069-2087
- ISSN
- 1475-9217 (ISSN-L)
- 1741-3168
- OpenAlex
- W3092593798
- Cited by
- 236
- References
- 42
- Citations by year
- 2026: 37 · 2025: 80 · 2024: 43 · 2023: 36 · 2022: 27 · 2021: 13
- FWCI
- 11.10
- Type
- journal article
- Language
- en
- Open access
- closed
- Licence
- journals.sagepub.com/page/policies/text-and-data-mining-license (tdm)
- Locations
- Structural Health Monitoring (publishedVersion)
- Funders
- China Scholarship Council (201906260157)
- Science and Technology Commission of Shanghai Municipality (19DZ1203004)
- National Natural Science Foundation of China (51878482, 51978508)
- National Key R&D Program of China (2017YFC1500605)
- Topics
- Physical Sciences › Engineering › Civil and Structural Engineering › Structural Health Monitoring Techniques
- Physical Sciences › Physics and Astronomy › Statistical and Nonlinear Physics › Model Reduction and Neural Networks
- Physical Sciences › Engineering › Civil and Structural Engineering › Infrastructure Maintenance and Monitoring
- Keywords
- missing data imputation, convolutional generative adversarial networks, sensor faults, modal parameter identification
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
Data sources
Bibliographic data: Semantic Scholar, OpenAlex, Crossref.
Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).
Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.