Root cause diagnosis for complex industrial process faults via spatiotemporal coalescent based time series prediction and optimized Granger causality

Wang, Sheng; Zhao, Qiang; Han, Yinghua; Wang, Jinkuan

doi:10.1016/j.chemolab.2022.104728

Cited by 9 publications

(7 citation statements)

References 34 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…When compared to conventional methodologies, this approach can yield more precise and dependable data, enabling better reservoir characterisation and production predictions. In Wang, S., et al's discussion [25], To precisely identify complex industrial process defects, the root cause diagnosis approach blends optimal Granger causality with spatiotemporal coalescent based time series prediction. This approach takes into account the interactions between various variables in the system to establish causation and identify the root cause of the issue, in addition to using past data to predict future behavior.…”

Section: Related Wordsmentioning

confidence: 99%

Prediction Analysis of Fault Mode Sets for High Permeability Distributed Energy Distribution Networks Based on Multivariate Data

Jianwei Ma

2024

jes

View full text Add to dashboard Cite

This technical abstract describes a prediction analysis methodology for fault mode sets in high-permeability distributed energy distribution networks (DEDNs). DEDNs are complex systems composed of multiple interconnected energy sources, storage units, and loads, which are managed and controlled by intelligent devices. These systems have become increasingly important in modern electricity networks due to their ability to integrate a high share of renewable and distributed energy resources. The proposed prediction analysis aims to identify and classify the various fault modes that can occur in high-permeability DEDNs based on multivariate data. It includes information about the system's operational parameters, such as voltage, current, and frequency, as well as data on weather conditions, load profiles, and the state of the network. The methodology involves collecting and pre-processing the data using suitable techniques, such as data filtering and noise removal.

show abstract

Section: Related Wordsmentioning

confidence: 99%

Prediction Analysis of Fault Mode Sets for High Permeability Distributed Energy Distribution Networks Based on Multivariate Data

Jianwei Ma

2024

jes

View full text Add to dashboard Cite

show abstract

“…The streaming data literature offers potentially valuable works on clustering [767], pre-processing [768], outlier treatment [769,770], and event prediction [771], although we were unable to identify mutual references in the analyzed literature. Causality analysis is an example of an application based on a time series that has gained prominence in the PSE field, as evidenced by numerous studies [580,[772][773][774][775][776][777][778][779][780][781][782]. This technique uses statistical tests, such as the Granger causality test, to determine whether a given time series is useful in predicting another.…”

Section: Cross-domain Integrationmentioning

confidence: 99%

Data-Driven Process Monitoring and Fault Diagnosis: A Comprehensive Survey

Melo,

Câmara,

Pinto

2024

Processes

View full text Add to dashboard Cite

This paper presents a comprehensive review of the historical development, the current state of the art, and prospects of data-driven approaches for industrial process monitoring. The subject covers a vast and diverse range of works, which are compiled and critically evaluated based on the different perspectives they provide. Data-driven modeling techniques are surveyed and categorized into two main groups: multivariate statistics and machine learning. Representative models, namely principal component analysis, partial least squares and artificial neural networks, are detailed in a didactic manner. Topics not typically covered by other reviews, such as process data exploration and treatment, software and benchmarks availability, and real-world industrial implementations, are thoroughly analyzed. Finally, future research perspectives are discussed, covering aspects related to system performance, the significance and usefulness of the approaches, and the development environment. This work aims to be a reference for practitioners and researchers navigating the extensive literature on data-driven industrial process monitoring.

show abstract

“…In root cause analysis [54], nearly all causal analysis methods are highly sensitive to the number of candidate variables analyzed. When more variables are involved, the results of the root cause analysis become less clear and precise.…”

Section: Anomaly Interpretationmentioning

confidence: 99%

“…Therefore, similar to root cause analysis methods in fault diagnosis [54], it is crucial to first filter anomaly variables. Anomalies in industrial systems typically originate from a single variable and gradually spread to other related variables.…”

Section: Anomaly Interpretationmentioning

confidence: 99%

“…Existing methods for selecting anomaly or fault variables generally involve selecting the TopK variables based on the cumulative fault contribution or anomaly scores during the fault or anomaly period. The value of k is often chosen based on experience or prior knowledge [54], or the mean value is directly used as a threshold [55] for selecting candidate variables. These methods either rely too heavily on experience and prior knowledge or may result in incorrect selections when there are significant deviations in the anomaly scores.…”

Section: Anomaly Interpretationmentioning

confidence: 99%

See 1 more Smart Citation

Anomaly detection model for large-scale industrial systems using transfer entropy and graph attention network

Liang,

Pi,

Zhang

2024

Meas. Sci. Technol.

View full text Add to dashboard Cite

Multivariate time series (MTS) anomaly detection is vital for ensuring the safety and reliability of large-scale industrial systems. However, existing deep learning methods often overlook complex interrelationships between different time series and the study of anomalies has been limited to detection. To address this, we propose an MTS anomaly detection model based on transfer entropy (TE) and graph attention network (GAT). In the graph construction module, by combining modified TE with automatic structure learning, we extract intricate relationships between features. In the prediction module, we modify the GAT to implement the dynamic attention mechanism and non-linear interaction between different features to improve the accuracy of model prediction. Finally, our model combines the modified TE with anomaly detection task, which can be used to provide interpretability for the detected anomalies using the constructed causal graph. Experimental results on both real and public datasets show that our approach outperforms the mainstream methods, in particular, achieving optimal results in terms of F1 scores and recall.

show abstract

Root cause diagnosis for complex industrial process faults via spatiotemporal coalescent based time series prediction and optimized Granger causality

Cited by 9 publications

References 34 publications

Prediction Analysis of Fault Mode Sets for High Permeability Distributed Energy Distribution Networks Based on Multivariate Data

Prediction Analysis of Fault Mode Sets for High Permeability Distributed Energy Distribution Networks Based on Multivariate Data

Data-Driven Process Monitoring and Fault Diagnosis: A Comprehensive Survey

Anomaly detection model for large-scale industrial systems using transfer entropy and graph attention network

Contact Info

Product

Resources

About