A novel anomaly detection method based on adaptive Mahalanobis-squared distance and one-class kNN rule for structural health monitoring under environmental effects

Sarmadi, Hassan; Karamodin, Abbas

doi:10.1016/j.ymssp.2019.106495

Cited by 165 publications

(120 citation statements)

References 37 publications

Supporting

Mentioning

108

Contrasting

Order By: Relevance

“…To finally compare the damage-sensitive features relevant to the undamaged state, namely the baseline, and to the current state, a distance metric must be introduced. The Mahalanobis distance is a statistical tool for computing the dissimilarity between two multivariate datasets, or matrices [12]. If the feature matrices X and Z are handled in the distance calculation, the procedure may result in being time-consuming and cumbersome, as they are high-dimensional features.…”

Section: Feature Classification By Mahalanobis Distance Metricmentioning

confidence: 99%

“…For this purpose, it is necessary to generate the training and test sets Tx and Tz from Bx and Bz in the training and inspection phases [12]. Subsequently, the mean vector (vx) and the covariance matrix (Cx) are computed for the training set Tx, so as to measure the dissimilarity of each vector (tz) of the matrix Tz from these components in the following form:…”

Section: Feature Classification By Mahalanobis Distance Metricmentioning

confidence: 99%

“…Figure 4 illustrates the result of early damage detection via the Mahalanobis distance, where the first 123 samples are related to the normal condition of the bridge and the second 123 samples belong to the damaged state; the distance values regarding samples 1-92 pertain to normal conditions in the training phase. These results have been obtained by handling each of the vectors (tx) of the training matrix in Equation (5), to estimate the threshold value as the horizontal dashed line in Figure 4, which is based on the generalized extreme value distribution and block maxima technique [10,12]. The distance quantities relevant to samples 93-123 refer to the same normal conditions, but used as validation data.…”

Section: Performance Validationmentioning

confidence: 99%

“…Furthermore, novelty detection based on unsupervised learning, which is to be contrasted with supervised learning [10,11], is an influential method for feature classification. Statistical distances [4,[12][13][14], clustering algorithms [15] and artificial neural networks [16] are popular tools for developing novelty detectors for SHM.…”

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

An Unsupervised Learning Approach for Early Damage Detection by Time Series Analysis and Deep Neural Network to Deal with Output-Only (Big) Data

Entezami

Sarmadi

Mariani

2020

7th International Electronic Conference on Sensors and Applications

Self Cite

View full text Add to dashboard Cite

Dealing with complex engineering problems characterized by Big Data, particularly in structural engineering, has recently received considerable attention due to its high societal importance. Data-driven structural health monitoring (SHM) methods aim at assessing the structural state and detecting any adverse change caused by damage, so as to guarantee structural safety and serviceability. These methods rely on statistical pattern recognition, which provides opportunities to implement a long-term SHM strategy by processing measured vibration data. However, the successful implementation of the data-driven SHM strategies when Big Data are to be processed is still a challenging issue, since the procedures of feature extraction and/or feature classification may end up being time-consuming and complex. To enhance the current damage detection procedures, in this work we propose an unsupervised learning method based on time series analysis, deep learning and the Mahalanobis distance metric for feature extraction, dimensionality reduction and classification. The main novelty of this strategy is the simultaneous dealing with the significant issue of Big Data analytics for damage detection, and distinguishing damage states from the undamaged one in an unsupervised learning manner. Large-scale datasets relevant to a cable-stayed bridge have been handled to validate the effectiveness of the proposed data-driven approach. Results have shown that the approach is highly successful in detecting early damage, even when Big Data are to be processed.

show abstract

Section: Feature Classification By Mahalanobis Distance Metricmentioning

confidence: 99%

Section: Feature Classification By Mahalanobis Distance Metricmentioning

confidence: 99%

Section: Performance Validationmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

An Unsupervised Learning Approach for Early Damage Detection by Time Series Analysis and Deep Neural Network to Deal with Output-Only (Big) Data

Entezami

Sarmadi

Mariani

2020

7th International Electronic Conference on Sensors and Applications

Self Cite

View full text Add to dashboard Cite

show abstract

“…The implementation of an unsupervised learning method for damage detection is generally carried out in the training and inspection phases 19 . During the training period, one attempts to learn a model via a large amount of the information (the training data) regarding the normal condition and obtain the outputs of the model of interest.…”

Section: Introductionmentioning

confidence: 99%

Ensemble learning‐based structural health monitoring by Mahalanobis distance metrics

Sarmadi

Entezami

Razavi³

et al. 2020

Struct Control Health Monit

Self Cite

View full text Add to dashboard Cite

Unsupervised machine and deep learning methods for structural damage detection: A comparative study

Wang

Cha

2022

Engineering Reports

View full text Add to dashboard Cite

While many structural damage detection methods have been developed in recent decades, few data‐driven methods in unsupervised learning mode have been developed to solve the practical difficulties in data acquisition for civil infrastructures in different scenarios. To address such a challenge, this article proposes a number of improved unsupervised novelty detection methods and conducts extensive comparative studies on a laboratory scale steel bridge to examine their performances of damage detection. The key concept behind unsupervised novelty detection in this article is that only normal data from undamaged/baseline structural scenarios are required to train statistical models with these methods. Then, these trained models are used to identify abnormal testing data from damaged scenarios. To detect structural damage in the form of loosening bolts in the steel bridge, four machine‐learning methods (i.e., K‐nearest neighbors method, Gaussian mixture models, one‐class support vector machines, density peaks‐based fast clustering method) and one deep learning method using a deep auto‐encoder are selected. Meanwhile, some modifications and improvements are made to enable these methods to detect structural damage in unsupervised novelty detection mode. In their comparative studies, the advantages and disadvantages of these methods are analyzed based on their results of structural damage detection.

show abstract

A novel anomaly detection method based on adaptive Mahalanobis-squared distance and one-class kNN rule for structural health monitoring under environmental effects

Cited by 165 publications

References 37 publications

An Unsupervised Learning Approach for Early Damage Detection by Time Series Analysis and Deep Neural Network to Deal with Output-Only (Big) Data

An Unsupervised Learning Approach for Early Damage Detection by Time Series Analysis and Deep Neural Network to Deal with Output-Only (Big) Data

Ensemble learning‐based structural health monitoring by Mahalanobis distance metrics

Unsupervised machine and deep learning methods for structural damage detection: A comparative study

Contact Info

Product

Resources

About