Detecting outliers in industrial systems using a hybrid ensemble scheme

Wang, Biao; Mao, Zhizhong

doi:10.1007/s00521-019-04307-5

Cited by 10 publications

(3 citation statements)

References 31 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Let S = {s (1) , s (2) , • • • , s (T) } ∈ N×T represent a training dataset of multivariate time series with timestamps produced by N sensors. At each time point s (t) ∈ R N , it represents the values of N sensors obtained at time t(t ≤ T).…”

Section: Problem Statementmentioning

confidence: 99%

“…Sensors are indispensable in these complex and large distributed systems, constantly capturing data on the operation of industrial equipment. This results in an immense amount of multivariate time series (MTS) data [1], such as the pressure, temperature, and fluid levels from chemical production machines, and the load, voltage, and current from sensors in power plants. However, the large scale and intricate structures of these MTS data, coupled with their dynamic nature, significantly heighten the risks of cyber-physical attacks.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Graph Transformer Network Incorporating Sparse Representation for Multivariate Time Series Anomaly Detection

Yang,

Zhang,

Zhang

et al. 2024

Electronics

View full text Add to dashboard Cite

Cyber–physical systems (CPSs) serve as the pivotal core of Internet of Things (IoT) infrastructures, such as smart grids and intelligent transportation, deploying interconnected sensing devices to monitor operating status. With increasing decentralization, the surge in sensor devices expands the potential vulnerability to cyber attacks. It is imperative to conduct anomaly detection research on the multivariate time series data that these sensors produce to bolster the security of distributed CPSs. However, the high dimensionality, absence of anomaly labels in real-world datasets, and intricate non-linear relationships among sensors present considerable challenges in formulating effective anomaly detection algorithms. Recent deep-learning methods have achieved progress in the field of anomaly detection. Yet, many methods either rely on statistical models that struggle to capture non-linear relationships or use conventional deep learning models like CNN and LSTM, which do not explicitly learn inter-variable correlations. In this study, we propose a novel unsupervised anomaly detection method that integrates Sparse Autoencoder with Graph Transformer network (SGTrans). SGTrans leverages Sparse Autoencoder for the dimensionality reduction and reconstruction of high-dimensional time series, thus extracting meaningful hidden representations. Then, the multivariate time series are mapped into a graph structure. We introduce a multi-head attention mechanism from Transformer into graph structure learning, constructing a Graph Transformer network forecasting module. This module performs attentive information propagation between long-distance sensor nodes and explicitly models the complex temporal dependencies among them to enhance the prediction of future behaviors. Extensive experiments and evaluations on three publicly available real-world datasets demonstrate the effectiveness of our approach.

show abstract

Section: Problem Statementmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Graph Transformer Network Incorporating Sparse Representation for Multivariate Time Series Anomaly Detection

Yang,

Zhang,

Zhang

et al. 2024

Electronics

View full text Add to dashboard Cite

show abstract

“…Accordingly, one-class classifiers are also referred to as data description techniques. During recent years, ensemble learning technique has been used in the domain of outlier detection [19][20][21]. In order to achieve more robust detecting performance, several ensemble methods developed in traditional classification has been used.…”

Section: Related Workmentioning

confidence: 99%

Enhancing the Prediction of Mach Number in Wind Tunnel With a Regression-Based Outlier Detection Framework

et al. 2021

Self Cite

View full text Add to dashboard Cite

Model predictive control (MPC) has been successfully applied in practical wind tunnel systems. As a critical component of MPC, the accurate prediction of Mach number becomes hence very significant. Real measurement data records often contain outliers or noisy data because of the harsh working conditions of practical systems. Such datasets are detrimental to most data-driven Mach number predictors. This paper focuses on improving Mach number prediction via a regression-based outlier detection framework. The development of a regression-based outlier detector for enhancing Mach number prediction is natural since only regression analysis needs to be implemented for both prediction and outlier detection. In contrast to existing regression-based detectors, we have taken into account three problems that may greatly impact the detecting performance. Specifically, we use Bagging technique to reduce the negative influence of unknown outliers during the training phase. Then, the notion of extreme value analysis is used to determine an appropriate threshold for the calculated outlier scores. A robust scaling method based on Hampel identifier is also used to alleviate the influence of outliers on data scaling. Several datasets stemming from a real-world wind tunnel are used in experiments to verify the effectiveness of our detector. The experimental results show that regression-based outlier detectors often perform better if the prior knowledge about the dependent variable is used. In addition, the proposed detector has outperformed other regression-based detectors.

show abstract

NaNOD: A natural neighbour-based outlier detection algorithm

Wahid

Annavarapu

2020

Neural Comput & Applic

View full text Add to dashboard Cite

Detecting outliers in industrial systems using a hybrid ensemble scheme

Cited by 10 publications

References 31 publications

Graph Transformer Network Incorporating Sparse Representation for Multivariate Time Series Anomaly Detection

Graph Transformer Network Incorporating Sparse Representation for Multivariate Time Series Anomaly Detection

Enhancing the Prediction of Mach Number in Wind Tunnel With a Regression-Based Outlier Detection Framework

NaNOD: A natural neighbour-based outlier detection algorithm

Contact Info

Product

Resources

About