Bidirectional deep recurrent neural networks for process fault classification

Chadha, Gavneet Singh; Panambilly, Ambarish; Schwung, Andreas; Ding, Steven X.

doi:10.1016/j.isatra.2020.07.011

Cited by 77 publications

(22 citation statements)

References 38 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Table 1 represents the Tennessee Eastman Process fault cases. Since the TE process dataset contains collected time-series sensor data, the data is prepared as time series sequences as discussed in [ 1 ] before the training.…”

Section: Resultsmentioning

confidence: 99%

“…To define the anomaly detection setting, we follow previous works [ 1 ] by dividing the fault classes into subgroups based on how challenging the faults are to detect. Accordingly, we divide the 21 faults into three subgroups: easy, medium, and hard-to-detect faults.…”

Section: Resultsmentioning

confidence: 99%

“…This unreliability can also lead to substantial financial consequences. However, modern production systems constitute complex interconnected behaviour, which renders the derivation of models through the first principle very difficult [ 1 ]. Hence, data-driven methods are an appealing alternative, particularly as a huge amount of data ranging from field level devices like sensors and actuators to manufacturing execution systems and enterprise resource planning systems are available through the Industrial Internet of Things [ 2 ].…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Deep Convolutional Clustering-Based Time Series Anomaly Detection

Chadha

Islam

Schwung

et al. 2021

Sensors

Self Cite

View full text Add to dashboard Cite

This paper presents a novel approach for anomaly detection in industrial processes. The system solely relies on unlabeled data and employs a 1D-convolutional neural network-based deep autoencoder architecture. As a core novelty, we split the autoencoder latent space in discriminative and reconstructive latent features and introduce an auxiliary loss based on k-means clustering for the discriminatory latent variables. We employ a Top-K clustering objective for separating the latent space, selecting the most discriminative features from the latent space. We use the approach to the benchmark Tennessee Eastman data set to prove its applicability. We provide different ablation studies and analyze the method concerning various downstream tasks, including anomaly detection, binary and multi-class classification. The obtained results show the potential of the approach to improve downstream tasks compared to standard autoencoder architectures.

show abstract

Section: Resultsmentioning

confidence: 99%

Section: Resultsmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Deep Convolutional Clustering-Based Time Series Anomaly Detection

Chadha

Islam

Schwung

et al. 2021

Sensors

Self Cite

View full text Add to dashboard Cite

show abstract

“…Fault-1,2,4,5,6,7 are all step-type ones which are considered by [36]- [37] to demonstrate the good performance of their fault diagnosis methods. Similarly, this paper employs the SOM (self-organizing map, results are from [37]), CCA-SOM (canonical correlation analysis and self-organizing map, results are from [37]) and B-LSTM (Bidirectional Long Short Term Memory, results are from [36]) to demonstrate the effectiveness of this SGCN-based fault diagnosis, and the diagnostic results of 7 classes are presented in Table 4. Most diagnosis accuracy of these 7 classes can be reached to above 99.2% except normal and fault-5.…”

Section: Case-3mentioning

confidence: 99%

Spatial-temporal associations representation and application for process monitoring using graph convolution neural network

Ren¹,

Yang²,

Liang³

et al. 2022

Preprint

View full text Add to dashboard Cite

Industrial process data reflects the dynamic changes of operation conditions, which mainly refer to the irregular changes in the dynamic associations between different variables in different time. And this related associations knowledge for process monitoring is often implicit in these dynamic monitoring data which always have richer operation condition information and have not been paid enough attention in current research. To this end, a new process monitoring method based on spatial-based graph convolution neural network (SGCN) is proposed to describe the characteristics of the dynamic associations which can be used to represent the operation status over time. Spatia-temporal graphs are firstly defined, which can be used to represent the characteristics of node attributes (dynamic edge features) dynamically changing with time. Then, the associations between monitoring variables at a certain time can be considered as the node attributes to define a snapshot of the static graph network at the certain time. Finally, the snapshot containing graph structure and node attributes is used as model inputs which are processed to implement graph classification by spatial-based convolution graph neural network with aggregate and readout steps. The feasibility and applicability of this proposed method are demonstrated by our experimental results of benchmark and practical case application.

show abstract

“…In our model, the bidirectional network starts with a spatial dropout layer, which performs dropout on entire feature maps rather than individual elements. Then, the output of this layer is fed to the bidirectional RNN (BiRNN) layer [32] based on the GRU, which connects two hidden layers (forward and backward) of opposite directions to the same output. After that, the output of the BiRNN layer is fed the global average pooling layer and global maximum pooling layer simultaneously, and the outputs of the two layers are combined to form new input to the next stage, as shown in Figure 2.…”

Section: Bidirectional Recurrent Neural Networkmentioning

confidence: 99%

A Multichannel Deep Learning Framework for Cyberbullying Detection on Social Media

Alotaibi

Razaque

2021

Electronics

View full text Add to dashboard Cite

Online social networks (OSNs) play an integral role in facilitating social interaction; however, these social networks increase antisocial behavior, such as cyberbullying, hate speech, and trolling. Aggression or hate speech that takes place through short message service (SMS) or the Internet (e.g., in social media platforms) is known as cyberbullying. Therefore, automatic detection utilizing natural language processing (NLP) is a necessary first step that helps prevent cyberbullying. This research proposes an automatic cyberbullying method to detect aggressive behavior using a consolidated deep learning model. This technique utilizes multichannel deep learning based on three models, namely, the bidirectional gated recurrent unit (BiGRU), transformer block, and convolutional neural network (CNN), to classify Twitter comments into two categories: aggressive and not aggressive. Three well-known hate speech datasets were combined to evaluate the performance of the proposed method. The proposed method achieved promising results. The accuracy of the proposed method was approximately 88%.

show abstract

Bidirectional deep recurrent neural networks for process fault classification

Cited by 77 publications

References 38 publications

Deep Convolutional Clustering-Based Time Series Anomaly Detection

Deep Convolutional Clustering-Based Time Series Anomaly Detection

Spatial-temporal associations representation and application for process monitoring using graph convolution neural network

A Multichannel Deep Learning Framework for Cyberbullying Detection on Social Media

Contact Info

Product

Resources

About