A combined canonical variate analysis and Fisher discriminant analysis (CVA–FDA) approach for fault diagnosis

Jiang, Benben; Zhu, Xiaoxiang; Huang, Dexian; Paulson, Joel A.; Braatz, Richard D.

doi:10.1016/j.compchemeng.2015.03.001

Cited by 93 publications

(37 citation statements)

References 24 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…CVA models the relationship between the past vector p b ( t ) and the future vector f b ( t ), and can maximize the correlation between p b ( t ) and f b ( t ), which will achieve optimal predictors . On the one hand, it tries to find one canonical vector x b ( t ) (also called the state vector), which is the linear combinations of p b ( t ); on the other hand, it tries to find the other canonical vector r b ( t ), which is the linear combinations of f b ( t ):

({, \begin{array}{c} x_{b} ((), t) = J_{b} p_{b} ((), t) \\ r_{b} ((), t) = L_{b} f_{b} ((), t) \end{array}),

where J b and L b are transformation matrices that maximize the correlations between the canonical variates in canonical vector x b ( t ) and those in canonical vector r b ( t ).…”

Section: Distributed Canonical Variate Analysismentioning

confidence: 99%

“…CVA models the relationship between the past vector p b (t) and the future vector f b (t), and can maximize the correlation between p b (t) and f b (t), which will achieve optimal predictors 22 . On the one hand, it tries to find one canonical vector x b (t) (also called the state vector), which is the linear combinations of p b (t); on the other hand, it tries to find the other canonical vector r b (t), which is the linear combinations of f b (t):…”

Section: Offline Model Developmentmentioning

confidence: 99%

See 1 more Smart Citation

Large‐scale dynamic process monitoring based on performance‐driven distributed canonical variate analysis

Liu

Song

Zhao

et al. 2020

Journal of Chemometrics

View full text Add to dashboard Cite

As a typical process monitoring method for the large‐scale industrial process, the distributed principal components analysis (DPCA) needs to be improved because of its rough selection for the variables in each subblock. Moreover, for DPCA, the process dynamic property is ignored and invalid fault diagnosis may occur. Therefore, a performance‐driven distributed canonical variate analysis (DCVA) is proposed. Firstly, with historical fault information, the genetic algorithm is utilized to select appropriate variables for each subblock; secondly, canonical variate analysis is introduced to capture the dynamic information for performance improvement; finally, a novel fault diagnosis method is developed for the DCVA model. Case studies on a numerical example and the Tennessee Eastman benchmark process demonstrate the effectiveness of the proposed model. Highlights A novel fault diagnosis approach based on the distributed CVA model is presented. The dynamic property of process data for each sub‐block is firstly captured by CVA. With the genetic algorithm, the historical fault information is utilized to select appropriate variables in each sub‐block. The superiority of the developed method is validated on the numerical example and the TE process.

show abstract

({, \begin{array}{c} x_{b} ((), t) = J_{b} p_{b} ((), t) \\ r_{b} ((), t) = L_{b} f_{b} ((), t) \end{array}),

where J b and L b are transformation matrices that maximize the correlations between the canonical variates in canonical vector x b ( t ) and those in canonical vector r b ( t ).…”

Section: Distributed Canonical Variate Analysismentioning

confidence: 99%

“…CVA models the relationship between the past vector p b (t) and the future vector f b (t), and can maximize the correlation between p b (t) and f b (t), which will achieve optimal predictors 22 . On the one hand, it tries to find one canonical vector x b (t) (also called the state vector), which is the linear combinations of p b (t); on the other hand, it tries to find the other canonical vector r b (t), which is the linear combinations of f b (t):…”

Section: Offline Model Developmentmentioning

confidence: 99%

Large‐scale dynamic process monitoring based on performance‐driven distributed canonical variate analysis

Liu

Song

Zhao

et al. 2020

Journal of Chemometrics

View full text Add to dashboard Cite

show abstract

“…By defining Γ = 1 N k −1 P A T S w,k P A , the FDA discriminant function is constructed as [11], [12]…”

Section: Fisher Discriminant Analysismentioning

confidence: 99%

“…2 simulates a real chemical process, which involves five units: reactor, condenser, compressor, separator and stripper. The TE process simulator was firstly built by Downs and Vogel [25] and has been widely used for validating various process monitoring and fault diagnosis techniques [4], [6], [10], [12], [26]. With the TE simulator, operation datasets for normal operation mode and 21 classes of fault modes have been generated.…”

Section: A Simulation Case Studymentioning

confidence: 99%

“…Chiang et al developed an FDA based industrial process fault classification method [10] and then proposed a dynamic FDA method by augmenting the observed vectors [11]. To deal with the serial correlation property of process data, Jiang et al [12] developed a CVA-FDA method which uses CVA to extract dynamic features for helping fault discrimination. For utilizing both labeled and unlabeled samples, Zhong et al [13] presented a semi-supervised FDA model which combines the objectives of PCA and FDA.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Statistics local fisher discriminant analysis for industrial process fault classification

Deng

Tian

Chen

et al. 2016

2016 UKACC 11th International Conference on Control (CONTROL)

View full text Add to dashboard Cite

In order to effectively identify industrial process faults, an improved Fisher discriminant analysis (FDA) method, referred to as the statistics local Fisher discriminant analysis (SLFDA), is proposed for fault classification. For mining statistics information hidden in process data, statistics pattern analysis is firstly applied to transform the original measured variables into the corresponding statistics, including second-order and higher-order ones. Furthermore, considering the local structure characteristics of fault data, local FDA (LFDA) is performed which computes the discriminant vectors by modifying the optimization objective with local weighting factor. Simulation results on the benchmark Tennessee Eastman process show that the proposed SLFDA has a better fault classification performance than the FDA and LFDA methods.

show abstract

Assessing the value of information of data‐centric activities in the chemical processing industry 4.0

Reis

Kenett

2018

AIChE Journal

View full text Add to dashboard Cite

The quality of information generated in data‐driven empirical studies is of central importance in Industry 4.0. However, despite the undeniable and widely accepted importance, not sufficient attention has been devoted to its rigorous assessment and analysis. Consequently, if information quality cannot be measured, it also cannot be improved, and therefore current efforts for extracting value from big data empirical studies and data collectors are exposed to the risk of generating limited findings and insights, leading to suboptimal solutions. In this article, we describe and apply a framework for evaluating, analyzing and improving the quality of information generated in empirical studies called InfoQ, in the context of the Chemical Processing Industry (CPI). This systematic framework can be used by anyone involved in data‐driven activities, irrespectively of the context and specific goals. The application of InfoQ framework to several case studies is described in detail, to illustrate its practical relevance. © 2018 American Institute of Chemical Engineers AIChE J, 64: 3868–3881, 2018

show abstract

A combined canonical variate analysis and Fisher discriminant analysis (CVA–FDA) approach for fault diagnosis

Cited by 93 publications

References 24 publications

Large‐scale dynamic process monitoring based on performance‐driven distributed canonical variate analysis

Large‐scale dynamic process monitoring based on performance‐driven distributed canonical variate analysis

Statistics local fisher discriminant analysis for industrial process fault classification

Assessing the value of information of data‐centric activities in the chemical processing industry 4.0

Contact Info

Product

Resources

About