Multimodal process monitoring based on variational Bayesian PCA and Kullback-Leibler divergence between mixture models

Cao, Yue; Jan, Nabil Magbool; Huang, Biao; Fang, Mengxiang; Wang, Yalin; Gui, Weihua

doi:10.1016/j.chemolab.2020.104230

Cited by 17 publications

(5 citation statements)

References 41 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Xie and Shi proposed a moving window GMM to consider both the multimode characteristics and dynamic characteristics [149]. Cao et al established a variational Bayesian PCA for each GMM to model multimodality, and KLD between a mixture of Gaussians were used as statistics, it was found that the sensitivity to small faults increased [150]. Multimode methods can provide a more effective feature extraction in each local mode.…”

Section: Multi-model Methodsmentioning

confidence: 99%

A Review on Data-Driven Process Monitoring Methods: Characterization and Mining of Industrial Data

Sun

2022

Processes

View full text Add to dashboard Cite

Safe and stable operation plays an important role in the chemical industry. Fault detection and diagnosis (FDD) make it possible to identify abnormal process deviations early and assist operators in taking proper action against fault propagation. After decades of development, data-driven process monitoring technologies have gradually attracted attention from process industries. Although many promising FDD methods have been proposed from both academia and industry, challenges remain due to the complex characteristics of industrial data. In this work, classical and recent research on data-driven process monitoring methods is reviewed from the perspective of characterizing and mining industrial data. The implementation framework of data-driven process monitoring methods is first introduced. State of art of process monitoring methods corresponding to common industrial data characteristics are then reviewed. Finally, the challenges and possible solutions for actual industrial applications are discussed.

show abstract

Section: Multi-model Methodsmentioning

confidence: 99%

A Review on Data-Driven Process Monitoring Methods: Characterization and Mining of Industrial Data

Sun

2022

Processes

View full text Add to dashboard Cite

show abstract

“…然而, 任何给定观测的过程工况通常是未知的, 它可能属于任何过程工况. 文献 [19] 提出了一种基于变分贝叶斯主成分分析和混合模型间 KL 散度的多工况过程异常监测方法, 用 KL 散度测量参考混合模型和监测混合模型之间关于每个过程工况的相异性, 利用贝叶斯推理将统计信息和控制限进行融合, 得到全局监测结果. 文献 [104] 提出了一种新的基于邻域的全局协调框架, 用于模型对齐和多工况过程异常监测, 通过将局部模型对齐到全局模型来涉及工况间的相关性.…”

Section: 基于混合建模的多工况平稳过程异常监测unclassified

Data-driven anomaly monitoring methods for multimode processes: overview and perspectives

Zhang¹,

Zhou²,

Chen³

et al. 2023

Sci. Sin.-Inf.

View full text Add to dashboard Cite

Industrial processes usually need to function under multiple conditions because of changes in raw materials, setting points, and the external environment. Anomaly monitoring methods for multimode processes effectively ensure the safe operation of industrial systems, and they are a research hotspot of current process monitoring technology. Herein, anomaly monitoring methods for multimode processes are summarized, including overall modeling, adaptive modeling, hybrid modeling, and multi-model modeling methods. Research progress of multimode stationary processes and multimode nonstationary processes is then reviewed, and the advantages and disadvantages of these methods are analyzed. Finally, considering the characteristics of the practical industrial systems, the open problems in this field and the future development direction are summarized.

show abstract

“…The reaction system is shown in Equation ( 29). In addition, the process includes 12 manipulated variables XMV (1)(2)(3)(4)(5)(6)(7)(8)(9)(10)(11)(12), 41 measured variables XMEAS (1-41) and 21 fault IDVs . Figure 6 shows the TE process flow chart.…”

Section: Te Processmentioning

confidence: 99%

“…Among them, multivariate statistical process monitoring (MSPM) methods do not require an accurate model of the production process; they only require historical data and online data to monitor the process and have been widely used in the field of industrial fault monitoring. [3][4][5] The most commonly used multivariate statistical methods are principal component analysis (PCA) [6][7][8][9] and independent component analysis (ICA). [10][11][12] However, both PCA and ICA can find global structural features of the data while ignoring local features, which leads to insufficient information extraction.…”

Section: Introductionmentioning

confidence: 99%

“…this paper, 22 measured variables XMEAS and 11 manipulated variables XMV(1)(2)(3)(4)(5)(6)(7)(8)(9)(10)(11) are selected as the monitoring variables, and the component G in stream 9, namely, XMEAS(35) is selected as the quality variable y. The training set contains 500 normal samples, and the test set contains 15 different faults sets, each consisting of 960 samples is used to verify the performance of the GNPER model.…”

mentioning

confidence: 99%

See 1 more Smart Citation

Quality‐relevant and process‐relevant fault monitoring based on GNPER and the fault quantification index for industrial processes

Mou

Zhao

2022

Can J Chem Eng

View full text Add to dashboard Cite

Traditional quality-relevant fault monitoring methods focus on extracting the relationship between the global structural features of the process and quality variables but ignore the local features. At the same time, they lack the quantification of quality-relevant faults. To solve these problems, a quality-relevant and process-relevant fault monitoring method and its fault quantification index based on global neighbourhood preserving embedding regression (GNPER) are proposed. First, by seeking the direction of maximum global vari-

show abstract

Multimodal process monitoring based on variational Bayesian PCA and Kullback-Leibler divergence between mixture models

Cited by 17 publications

References 41 publications

A Review on Data-Driven Process Monitoring Methods: Characterization and Mining of Industrial Data

A Review on Data-Driven Process Monitoring Methods: Characterization and Mining of Industrial Data

Data-driven anomaly monitoring methods for multimode processes: overview and perspectives

Quality‐relevant and process‐relevant fault monitoring based on GNPER and the fault quantification index for industrial processes

Contact Info

Product

Resources

About