An angle-based subspace anomaly detection approach to high-dimensional data: With an application to industrial fault detection

Zhang, Liangwei; Lin, Jing; Karim, Ramin

doi:10.1016/j.ress.2015.05.025

Cited by 52 publications

(16 citation statements)

References 40 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…A local outlier score is then computed for these points indicating whether it is abnormal or not. More theoretical discussions to this approach can be referred to [30].…”

mentioning

confidence: 99%

Sliding Window-Based Fault Detection From High-Dimensional Data Streams

Zhang

Lin

Karim

2016

IEEE Trans. Syst. Man Cybern, Syst.

Self Cite

View full text Add to dashboard Cite

Abstract-High-dimensional data streams are becoming increasingly ubiquitous in industrial systems. Efficient detection of system faults from these data can ensure the reliability and safety of the system. The difficulties brought about by high dimensionality and data streams are mainly the "curse of dimensionality" and concept drifting, and one current challenge is to simultaneously address them. To this purpose, this paper presents an approach to fault detection from nonstationary highdimensional data streams. An angle-based subspace anomaly detection approach is proposed to detect low-dimensional subspace faults from high-dimensional datasets. Specifically, it selects fault-relevant subspaces by evaluating vectorial angles and computes the local outlier-ness of an object in its subspace projection. Based on the sliding window strategy, the approach is further extended to an online mode that can continuously monitor system states. To validate the proposed algorithm, we compared it with the local outlier factor-based approaches on artificial datasets and found the algorithm displayed superior accuracy. The results of the experiment demonstrated the efficacy of the proposed algorithm. They also indicated that the algorithm has the ability to discriminate low-dimensional subspace faults from normal samples in high-dimensional spaces and can be adaptive to the time-varying behavior of the monitored system. The online subspace learning algorithm for fault detection would be the main contribution of this paper.

show abstract

“…A local outlier score is then computed for these points indicating whether it is abnormal or not. More theoretical discussions to this approach can be referred to [30].…”

mentioning

confidence: 99%

Sliding Window-Based Fault Detection From High-Dimensional Data Streams

Zhang

Lin

Karim

2016

IEEE Trans. Syst. Man Cybern, Syst.

Self Cite

View full text Add to dashboard Cite

show abstract

“…Zhangn et al, [21] proposed the approach for increasing the accuracy of the traditional technique of the outlier detection. It focused on the detection accuracy in the high dimensional data.…”

Section: Literature Surveymentioning

confidence: 99%

“…The object with high local outlier factor is termed as outlier and the objects having low local outlier factor are considered to be normal. The high local outlier filter depicts the high probability of being outlier [21].…”

Section: Density Based Outlier Detectionmentioning

confidence: 99%

Comparative Study of Outlier Detection Algorithms

Kaur¹,

Garg²

2016

IJCA

View full text Add to dashboard Cite

As the dimension of the data is increasing day by day, outlier detection is emerging as one of the active area of research. Finding of the outliers from large data sets is the main problem. Outlier is considered as the pattern that is different from the rest of the patterns present in the data set. The detection of the outlier in the data set is an important process as it helps in acquiring the useful information that further helps in the data analysis. Various algorithms have been proposed till date for the detection of the outliers. This paper covers a study of various outlier detection algorithms like Statistical based outlier detection, Depth based outlier detection, Clustering based technique, Density based outlier detection etc. Comparison study of these outlier detection methods is done to find out which of the outlier detection algorithms are more applicable on high dimensional data.

show abstract

“…Signal-based methods require thorough analysis and a priori knowledge on fault mechanism. In addition, the manually extracted feature has a limitation in terms of application, that is, it is only suitable for specific diagnosis issues, thus limiting the application in complex chemical systems [8]. …”

Section: Introductionmentioning

confidence: 99%

“…This method is also called intelligent fault diagnosis method, where artificial intelligence techniques are combined [9]. This method attempts to acquire underlying knowledge from large amounts of empirical data through model learning and is more desirable than other methods [8]. As representatives, artificial neural network (ANN), support vector machine (SVM), and multi-layer perceptron (MLP) have been applied successfully in the field of fault diagnosis [10,11,12,13].…”

Section: Introductionmentioning

confidence: 99%

Fault Diagnosis Based on Chemical Sensor Data with an Active Deep Neural Network

Jiang

et al. 2016

Sensors

View full text Add to dashboard Cite

Big sensor data provide significant potential for chemical fault diagnosis, which involves the baseline values of security, stability and reliability in chemical processes. A deep neural network (DNN) with novel active learning for inducing chemical fault diagnosis is presented in this study. It is a method using large amount of chemical sensor data, which is a combination of deep learning and active learning criterion to target the difficulty of consecutive fault diagnosis. DNN with deep architectures, instead of shallow ones, could be developed through deep learning to learn a suitable feature representation from raw sensor data in an unsupervised manner using stacked denoising auto-encoder (SDAE) and work through a layer-by-layer successive learning process. The features are added to the top Softmax regression layer to construct the discriminative fault characteristics for diagnosis in a supervised manner. Considering the expensive and time consuming labeling of sensor data in chemical applications, in contrast to the available methods, we employ a novel active learning criterion for the particularity of chemical processes, which is a combination of Best vs. Second Best criterion (BvSB) and a Lowest False Positive criterion (LFP), for further fine-tuning of diagnosis model in an active manner rather than passive manner. That is, we allow models to rank the most informative sensor data to be labeled for updating the DNN parameters during the interaction phase. The effectiveness of the proposed method is validated in two well-known industrial datasets. Results indicate that the proposed method can obtain superior diagnosis accuracy and provide significant performance improvement in accuracy and false positive rate with less labeled chemical sensor data by further active learning compared with existing methods.

show abstract

An angle-based subspace anomaly detection approach to high-dimensional data: With an application to industrial fault detection

Cited by 52 publications

References 40 publications

Sliding Window-Based Fault Detection From High-Dimensional Data Streams

Sliding Window-Based Fault Detection From High-Dimensional Data Streams

Comparative Study of Outlier Detection Algorithms

Fault Diagnosis Based on Chemical Sensor Data with an Active Deep Neural Network

Contact Info

Product

Resources

About