Condition-Driven Data Analytics and Monitoring for Wide-Range Nonstationary and Transient Continuous Processes

Zhao, Chunhui; Chen, Junhao; Hua, Jing

doi:10.1109/tase.2020.3010536

Cited by 55 publications

(16 citation statements)

References 36 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…Our future research direction is how to use the LVs extracted by ISFA to establish dynamic monitoring statistics and realize the use of ISFA to monitor dynamic and multimode processes. 27 …”

Section: Discussionmentioning

confidence: 99%

Fault Detection of Non-Gaussian and Nonlinear Processes Based on Independent Slow Feature Analysis

Zhou

Wen

et al. 2022

ACS Omega

View full text Add to dashboard Cite

Independent component analysis (ICA) is an excellent latent variables (LVs) extraction method that can maximize the non-Gaussianity between LVs to extract statistically independent latent variables and which has been widely used in multivariate statistical process monitoring (MSPM). The underlying assumption of ICA is that the observation data are composed of linear combinations of LVs that are statistically independent. However, the assumption is invalid because the observation data are always derived from the nonlinear mixture of LVs due to the nonlinear characteristic in industrial processes. Under this circumstance, the ICA-based fault detection is unable to provide accurate detection for specific faults of industrial processes. Since the observation data come from the nonlinear mixing of LVs, this makes the observation data change faster than the intrinsic LVs on the time scale. The temporal slowness can be regarded as an additional criterion in the extraction of LVs. The slow feature analysis (SFA) derived from the temporal slowness has received extensive attention and application in MSPM in recent years. Simultaneously, the temporal slowness is expected to make up for the problem that the LVs extracted by ICA have difficulty accurately describing the characteristics of the process. To solve the above problems, this work proposes to monitor non-Gaussian and nonlinear processes using the independent slow feature analysis (ISFA) that combines statistical independence and temporal slowness in extracting the LVs. When the observation data are composed of a nonlinear mixture of LVs, the extracted LVs of ISFA can describe the characteristics of the processes better than ICA, thereby improving the accuracy of fault detection for the non-Gaussian and nonlinear processes. The superiority of the proposed method is verified by a numerical example design and the Tennessee–Eastman process.

show abstract

Section: Discussionmentioning

confidence: 99%

Fault Detection of Non-Gaussian and Nonlinear Processes Based on Independent Slow Feature Analysis

Zhou

Wen

et al. 2022

ACS Omega

View full text Add to dashboard Cite

show abstract

“…If the process is multimodal with shifted operating conditions, different conditions can be distinguished first. Several studies , have been reported to distinguish different conditions based on which the multimode can be decomposed into multiple single-mode treatments.…”

Section: Methodsmentioning

confidence: 99%

Single Model-Based Analysis of Relative Causal Changes for Root-Cause Diagnosis in Complex Industrial Processes

Tian

Zhao

2021

Ind. Eng. Chem. Res.

Self Cite

View full text Add to dashboard Cite

Traditional root-cause diagnosis methods analyze different faults independently, suffering from repetitive modeling efforts for each fault. Besides, they are disturbed by significant redundancy and interference relations, being both time and effort consuming in online applications. In this work, a novel strategy is proposed to diagnose root cause by analyzing the changes in causal effects between the normal status and fault statuses, which allows a single causal model to diagnose different faults. The enhanced cross mapping (ECM) is proposed to carry out the analysis of the causal effects, which could detect precise causality and handle interference relations in complex industrial processes. In our strategy, a causal model is constructed from coarse-grained to fine-grained via the ECM to remove interference relations and achieve a concise structure. Besides, causal intensity and direction information of normal status are comprehensively described in the model. By updating causal effects under fault statuses, the relative causal changes from normal to fault can be analyzed and the significant changes are considered responsible for the fault propagation. Consequently, the fault propagation path is highlighted from the causal model and the root cause can be identified along the path. The feasibility of the proposed strategy is demonstrated using both simulation and practical industrial case studies.

show abstract

“…Thus, the processes share certain common feature information that can be transferred from one process to another process. In order to better understand and control those processes, establishing each process model based on a multiprocess joint modeling method timely, accurately, and comprehensively is very important. , Different from the traditional modeling methods focused on one process only, the multiprocess joint modeling method captures the useful common information among multiple processes to reduce the modeling cost and improve the generalization performance of each process. , Therefore, it is desirable to investigate a multiprocess joint modeling method that synthesizes the common information among multiprocess to model those processes accurately and simultaneously.…”

Section: Introductionmentioning

confidence: 99%

A Multiprocess Joint Modeling Method for Performance Prediction of Nonlinear Industrial Processes Based on Multitask Least Squares Support Vector Machine

Chu

Dai

et al. 2022

Ind. Eng. Chem. Res.

View full text Add to dashboard Cite

Developing the models of multiple processes rapidly and accurately is very important for process control and optimization in industrial production. This paper proposes a multiprocess joint modeling method to establish the models of multiple industrial processes simultaneously. Under the assumption that the related processes shared common information, this method views the process model as the combination of the common feature model and the special feature model. Through the full mining and using the shared information, the proposed method can establish models of multiple processes simultaneously and improve the accuracy of each process model. In particular, an improved multitask least squares support vector machine algorithm, which assumes that all the learning processes share the common feature model, is proposed to train multiprocess models simultaneously. Experiments developing the models of multiple centrifugal compressors are conducted, and the simulation experiment results show that the proposed method can realize joint modeling of multiple similar processes, improve modeling efficiency, and reduce modeling costs.

show abstract

Condition-Driven Data Analytics and Monitoring for Wide-Range Nonstationary and Transient Continuous Processes

Cited by 55 publications

References 36 publications

Fault Detection of Non-Gaussian and Nonlinear Processes Based on Independent Slow Feature Analysis

Fault Detection of Non-Gaussian and Nonlinear Processes Based on Independent Slow Feature Analysis

Single Model-Based Analysis of Relative Causal Changes for Root-Cause Diagnosis in Complex Industrial Processes

A Multiprocess Joint Modeling Method for Performance Prediction of Nonlinear Industrial Processes Based on Multitask Least Squares Support Vector Machine

Contact Info

Product

Resources

About