Automated Structural Damage Detection Using One-Class Machine Learning

Long, James; Büyüköztürk, Oral

doi:10.1007/978-3-319-04546-7_14

Cited by 23 publications

(15 citation statements)

References 9 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…This current study has only identified the resonant frequency from the antenna tower. If the system were to be used for structural health monitoring possible methods for monitoring changes in the displacement signal include statistical pattern recognition techniques (Sohn et al 2001), one-class machine learning methods (Long and Buyukozturk 2014), analysis of nonlinear features (Mohammadi Ghazi and Büyüköztürk 2016), or other damage detection algorithms. Without the need for physical access to instrument a structure, cameras can more easily collect data from structures that might otherwise be difficult or time consuming to instrument.…”

Section: Discussionmentioning

confidence: 99%

Video Camera–Based Vibration Measurement for Civil Infrastructure Applications

Chen

Davis

Wadhwa

et al. 2017

J. Infrastruct. Syst.

166

View full text Add to dashboard Cite

Visual testing, as one of the oldest methods for nondestructive testing (NDT), plays a large role in the inspection of civil infrastructure. As NDT has evolved, more quantitative techniques have emerged such as vibration analysis. New computer vision techniques for analyzing the small motions in videos, collectively called motion magnification, have been recently developed, allowing quantitative measurement of the vibration behavior of structures from videos. Video cameras offer the benefit of long range measurement and can collect a large amount of data at once because each pixel is effectively a sensor. This paper presents a video camera-based vibration measurement methodology for civil infrastructure. As a proof of concept, measurements are made of an antenna tower on top of the Green Building on the campus of the Massachusetts Institute of Technology (MIT) from a distance of over 175 m, and the resonant frequency of the antenna tower on the roof is identified with an amplitude of 0.21 mm, which was less than 1=170th of a pixel. Methods for improving the noise floor of the measurement are discussed, especially for motion compensation and the effects of video downsampling, and suggestions are given for implementing the methodology into a structural health monitoring (SHM) scheme for existing and new structures.

show abstract

Section: Discussionmentioning

confidence: 99%

Video Camera–Based Vibration Measurement for Civil Infrastructure Applications

Chen

Davis

Wadhwa

et al. 2017

J. Infrastruct. Syst.

166

View full text Add to dashboard Cite

show abstract

“…Several kernel functions have been used in SVM such as Gaussian and polynomial kernels. However, Gaussian kernel defined in Equation has gained much popularity in the area of machine learning, and it showed a high potential to be an appropriate setting for OCSVM . This kernel function has a user defined parameter denoted by σ , which determines the width of the Gaussian kernel and can hysterically affect the performance of the OCSVM.…”

Section: Proposed Methodsmentioning

confidence: 99%

“…However, Gaussian kernel defined in Equation (11) has gained much popularity in the area of machine learning, and it showed a high potential to be an appropriate setting for OCSVM. [19][20][21] This kernel function has a user defined parameter denoted by σ, which determines the width of the Gaussian kernel and can hysterically affect the performance of the OCSVM. This parameter may yield to overfit the model if it receives small values and underfit in case of large values.…”

Section: Self-tuning: Gaussian Kernel Parametersmentioning

confidence: 99%

Frequency domain decomposition-based multisensor data fusion for assessment of progressive damage in structures

Alamdari

Anaissi

Khoa

et al. 2018

Struct Control Health Monit

View full text Add to dashboard Cite

Summary In this paper, we focused on the development and verification of a solid and robust framework for structural condition assessment of real‐life structures using measured vibration responses, with the presence of multiple progressive damages occurring within the inspected structures. A self‐tuning learning method for structural condition assessment was proposed. Damage sensitive features were extracted using a frequency domain decomposition (FDD) approach to fuse all the measured responses, followed by random projection algorithm for dimensionality reduction. An automatic parameter selection method called Appropriate Distance to the Enclosing Surface (ADES) was used for tuning the classifier parameter. The effect of operational conditions on the robustness of the proposed method was also investigated, and it was realized that application of FDD to extract damage sensitive feature reduces the variation in the results. Promising results in the assessment of damage were obtained based on two comprehensive case studies, which included single and multiple damage scenarios. The contributions of the work are threefold. First, through two comprehensive case studies, we demonstrate that the frequency‐based feature from a single sensor might not be adequate enough to detect the progress of damage, even if the sensor is in the vicinity of damage. Second, we show that data fusion using FDD can reliably assess the severity of damage, and finally, we propose a new automated approach for tuning the classifier parameter.

show abstract

“…Large values of σ underfit the data, resulting in an inability to detect non‐trivial patterns. Long and Buyukozturk investigated three potential methods for automatic selection of this parameter, and found the method developed by Khazai et al to give good results. This method is also used for this paper.…”

Section: Statistical Pattern Recognitionmentioning

confidence: 99%

Decentralised one-class kernel classification-based damage detection and localisation

Long

Büyüköztürk

2016

Struct. Control Health Monit.

Self Cite

View full text Add to dashboard Cite

Summary In this paper, a data‐based damage detection algorithm that uses a novel one‐class kernel classifier for detection and localisation of damage is presented. The demands of wireless sensing are carefully considered in the development of this fully decentralised and automated methodology. The one‐class kernel classifier proposed in this paper is trained through a faster and simpler to implement iterative procedure than other kernel classification methods, while retaining the same advantages over parametric methods, making it especially attractive for embedded damage detection. Acceleration time series at each sensor location are processed into autoregressive and continuous wavelet transform‐based damage‐sensitive features. Baseline values of these features are used to train the classifier, which can then classify features from new tests as damaged or undamaged, as well as outputting a localisation index, which can be used to identify the location of damage in the structure. This methodology is evaluated using acceleration data taken from a steel‐frame laboratory structure under various damage scenarios. A number of parametric studies are also conducted to investigate the effect of sampling frequency and baseline data sample size. Copyright © 2016 John Wiley & Sons, Ltd.

show abstract

Automated Structural Damage Detection Using One-Class Machine Learning

Cited by 23 publications

References 9 publications

Video Camera–Based Vibration Measurement for Civil Infrastructure Applications

Video Camera–Based Vibration Measurement for Civil Infrastructure Applications

Frequency domain decomposition-based multisensor data fusion for assessment of progressive damage in structures

Decentralised one-class kernel classification-based damage detection and localisation

Contact Info

Product

Resources

About