CNN and Convolutional Autoencoder (CAE) based real-time sensor fault detection, localization, and correction

Jana, Debasish; Patil, Jayant; Herkal, Sudheendra; Nagarajaiah, Satish; Dueñas‐Osorio, Leonardo

doi:10.1016/j.ymssp.2021.108723

Cited by 96 publications

(33 citation statements)

References 45 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Hughes et al 30 presented a risk‐based solution of active learning by leveraging probabilistic classifiers for determining the health of structures. Jana et al 31 utilized deep learning models to detect fault in sensor data and locate the faulty sensor and consequently reconstructed the correct sensor data in real‐time.…”

Section: Related Workmentioning

confidence: 99%

Uncertainty‐aware convolutional neural network for explainable artificial intelligence‐assisted disaster damage assessment

Cheng

Behzadan

Noshadravan

2022

Structural Contr & Hlth

View full text Add to dashboard Cite

Accurate damage assessment is a critical step in post-disaster risk assessment, mitigation, and recovery. Current practices performed by experts and reconnaissance teams in the form of field evaluation require considerable time and resources. Recent advances in remote sensing imagery, artificial intelligence (AI), and computer vision have enhanced automated and rapid disaster damage assessment. Recent literature has shown promising progress in AI-assisted aerial damage assessment. However, accounting for the uncertainty in the outcome for improved quantification of confidence and enhanced model explainability for human decision-makers remains one of the key challenges.Overlooking uncertainty can lead to erroneous decisions, especially in highlyconsequential tasks such as damage assessment. The aim of this study is to develop uncertainty-aware deep learning models for the assessment of postdisaster damage using aerial imaging. Within the framework of variational Bayesian inference, Monte Carlo dropout sampling technique is used to propagate epistemic uncertainty in model predictions. With this stochastic setting, the model produces damage prediction labels with softmax as random variables, which helps quantify confidence in the model outcome using appropriate measures of uncertainty. Two networks are implemented and trained separately on two different disaster damage datasets consisting of unmanned aerial vehicle building footage as well as satellite-captured post-disaster imagery. The first network attains 59.4% accuracy in building classification, and the second network gives an accuracy of 55.1%. Results from uncertainty analysis, model confidence quantification, and analyzing model attention zone can lead to more explainable and risk-informed automated damage assessment outcomes using AI technology.

show abstract

Section: Related Workmentioning

confidence: 99%

Uncertainty‐aware convolutional neural network for explainable artificial intelligence‐assisted disaster damage assessment

Cheng

Behzadan

Noshadravan

2022

Structural Contr & Hlth

View full text Add to dashboard Cite

show abstract

“…These contact‐based sensors may malfunction when exposed to harsh weather conditions. This type of severe weather can be critical to any structure, and sensor malfunctioning would measure erroneous structural response, which could lead to expensive maintenance and sensor fault detection 23 . To circumvent these scenarios, adopting the non‐contact vision‐based measurement can be useful and less expensive.…”

Section: Introductionmentioning

confidence: 99%

Computer vision‐based real‐time cable tension estimation algorithm using complexity pursuit from video and its application in Fred‐Hartman cable‐stayed bridge

Jana

Nagarajaiah

Yang

2022

Structural Contr & Hlth

Self Cite

View full text Add to dashboard Cite

Summary Real‐time health monitoring of stay cables in cable‐stayed bridges is necessary for timely maintenance and to avoid unforeseen fatigue damage due to vortex‐induced vibration—mainly due to combination rain and wind‐related dynamic loads. Conventional contact‐based sensors may often malfunction in harsh weather conditions and are expensive to install and maintain. Therefore, recently, the usage of non‐contact camera‐based measurement is burgeoning in the domain of structural sensing. Non‐contact video‐based sensing provides a higher spatial resolution compared to conventional sensors along with a lower cost. Therefore, in this paper, we present a framework that uses video‐based measurement as multiple sensors to reduce the estimation error in determining the real‐time cable tension. First, we calculate the vibration response using the phase‐based motion estimation algorithm for various locations of interest. We then intuitively fuse the data from all the locations to estimate the real‐time frequency variation using a blind source separation (BSS) technique named complexity pursuit (CP). Finally, the real‐time stay‐cable tension is calculated from the real‐time frequency history using the taut‐string theory. The proposed algorithm is applied to Fred‐Hartman cable‐stayed bridge in Houston, Texas. The algorithm is validated using actual tension in the cable. We also show that the estimation error in the proposed sliding window‐based CP framework is considerably lesser than the conventional real‐time tension estimation technique using Short‐time Fourier Transform (STFT). The accurate estimation of stay‐cable tension from the video‐based measurement shows the significant potential of the proposed framework in the domain of structural health monitoring.

show abstract

“…Arul and Kareem [2] presented to use a relatively new time-series representation named "Shapelet Transform" in combination with a Random Forest classifier to autonomously identify anomalies in SHM data. Jana et al [20] proposed CNN for detecting the presence of a sensor fault and convolutional autoencoder to reconstruct sensor data based on its identified type. The proposed model was tested using simulated and experimental datasets to demonstrate its performance.…”

Section: Introductionmentioning

confidence: 99%

Multiclass anomaly detection in imbalanced structural health monitoring data using convolutional neural network

Zhao

Sadhu

Capretz

2022

J Infrastruct Preserv Resil

View full text Add to dashboard Cite

Structural health monitoring (SHM) system aims to monitor the in-service condition of civil infrastructures, incorporate proactive maintenance, and avoid potential safety risks. An SHM system involves the collection of large amounts of data and data transmission. However, due to the normal aging of sensors, exposure to outdoor weather conditions, accidental incidences, and various operational factors, sensors installed on civil infrastructures can get malfunctioned. A malfunctioned sensor induces significant multiclass anomalies in measured SHM data, requiring robust anomaly detection techniques as an essential data cleaning process. Moreover, civil infrastructure often has imbalanced anomaly data where most of the SHM data remain biased to a certain type of anomalies. This imbalanced time-series data causes significant challenges to the existing anomaly detection methods. Without proper data cleaning processes, the SHM technology does not provide useful insights even if advanced damage diagnostic techniques are applied. This paper proposes a hyperparameter-tuned convolutional neural network (CNN) for multiclass imbalanced anomaly detection (CNN-MIAD) modelling. The hyperparameters of the proposed model are tuned through a random search algorithm to optimize the performance. The effect of balancing the database is considered by augmenting the dataset. The proposed CNN-MIAD model is demonstrated with a multiclass time-series of anomaly data obtained from a real-life cable-stayed bridge under various cases of data imbalances. The study concludes that balancing the database with a time shift window to increase the database has generated the optimum results, with an overall accuracy of 97.74%.

show abstract

CNN and Convolutional Autoencoder (CAE) based real-time sensor fault detection, localization, and correction

Cited by 96 publications

References 45 publications

Uncertainty‐aware convolutional neural network for explainable artificial intelligence‐assisted disaster damage assessment

Uncertainty‐aware convolutional neural network for explainable artificial intelligence‐assisted disaster damage assessment

Computer vision‐based real‐time cable tension estimation algorithm using complexity pursuit from video and its application in Fred‐Hartman cable‐stayed bridge

Multiclass anomaly detection in imbalanced structural health monitoring data using convolutional neural network

Contact Info

Product

Resources

About