Batch Process Monitoring with Tensor Global–Local Structure Analysis

Luo, Lei; Bao, Shiyi; Gao, Zengliang; Yuan, Jun

doi:10.1021/ie402355f

Cited by 35 publications

(55 citation statements)

References 41 publications

(110 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…If the calculated fault statistic is J and the fault threshold is J th then f = 0 denotes a no-fault condition and f ≠ 0 denotes a fault condition. Therefore, the fault detection rate (FDR) and false alarm rate (FAR) can be defined as follows [38,[43][44][45][46][47][48] :…”

Section: Fault Detection and Diagnosis Using Pca And Spcamentioning

confidence: 99%

“…Fault diagnosis is mainly carried out by assessing the contribution of different variables to the SPE or to the relevant PC. [41,48] The diagnosis of faulty samples for the CW actuator fault simulated in the CSTH system using the FDR-FAR SPCA and four other methods (ie, PCA, IS, AV, and NL) using each of these techniques is described below.…”

Section: Fault Diagnosismentioning

confidence: 99%

See 1 more Smart Citation

A new criterion for selection of non‐zero loadings for sparse principal component analysis (SPCA)

2021

View full text Add to dashboard Cite

Sparse principal component analysis (SPCA) has recently emerged as an approach aimed at producing compact principal component loadings by suppressing spurious values and thus overcoming some limitations of the traditional principal component analysis (PCA). This paper proposes a fault detection and diagnosis (FDD) method based on SPCA; in this approach, the number of non-zero loadings (NNZL) of SPCAs is selected based on both the false alarm rate (FAR) and the fault detection rate (FDR). The criterion is to have lower FAR and higher FDR. This new feature makes SPCA better suited for FDD, which is demonstrated by comparing its performance with that of three other methods for finding loadings. The overall FDD performances of both PCA and SPCA-based techniques are illustrated using the benchmark continuous stirred tank heater (CSTH) process. The results show that the PCs derived based on the proposed criterion has a better fault diagnosis ability.

show abstract

Section: Fault Detection and Diagnosis Using Pca And Spcamentioning

confidence: 99%

Section: Fault Diagnosismentioning

confidence: 99%

A new criterion for selection of non‐zero loadings for sparse principal component analysis (SPCA)

2021

View full text Add to dashboard Cite

show abstract

“…The method of tensor analysis does not unfold the three-dimension data, but directly models the three-dimension data, so that the internal structure of the data can not be destroyed, and more useful information for monitoring can be saved. Luo et al [11] proposed a global and local structure analysis algorithm based on tensor to avoid the loss of information due to the expansion of three-dimension data. Zhao and Hui [12] proposed a temporal extension global-local neighborhood preserving embedding based on tensor factorization algorithm, which directly dealt with three-dimension data and extracted local and global feature information, which could get fully extracted fault information.…”

Section: Introductionmentioning

confidence: 99%

Markov Chain Neighborhood Sparse Preserving Graph Embedding Based on Tensor Factorization for Batch Process Monitoring

Zhao

Mou

2021

IEEE Access

View full text Add to dashboard Cite

If the three-dimension data of batch process are unfolded the two-dimension data, some important information would lose, and outliers such as noise would lead to poor monitoring results. Therefore, a Markov chain neighborhood sparse preserving graph embedding algorithm based on tensor factorization (TMNSPGE) is proposed. Firstly, tensor factorization is used to directly process the three-dimension data in batch process, which can avoid the information loss. Secondly, by using the neighborhood preserving embedding algorithm and sparse manifold coding, the local linear relationship and local sparse manifold structure of data are preserved. On this basis, Markov chain analysis is introduced to construct a similar graph to make the data after dimensionality reduction have a certain probability interpretation. Finally, the statistics and control limits are determined to realize process monitoring. Numerical example and penicillin fermentation simulation process prove the effectiveness of TMNSPGE algorithm in batch process monitoring. INDEX TERMS Batch process monitoring, finite Markov chain, graph embedding, neighborhood preserving embedding, sparse representation, tensor factorization.

show abstract

“…Initially, tensor‐based methods were applied to face recognition, such as tensor PCA (TPCA) and tensor linear discriminant analysis (TLDA) . Luo et al proposed a tensor global‐local structure analysis (TGLSA) algorithm to process 3‐diamensional data of the batch process, which considered both the global and local characteristics of the data. Rong et al designed a tensor locality preserving discriminant analysis (TLPDA) for dynamic fault diagnosis, which used augmented process data matrices as a sample.…”

Section: Introductionmentioning

confidence: 99%

Tensor sequence component analysis for fault detection in dynamic process

et al. 2019

View full text Add to dashboard Cite

For dynamic processes, using sequence information to augment the data can improve fault detection performance. Traditional approaches transform raw data into augmented vectors, which leads to losses in structural information in the variables and increases the data dimension. This paper proposes a novel data dimension reduction algorithm called tensor sequence component analysis (TSCA) and applies it to dynamic process fault detection. The algorithm extends each sample into a matrix comprising current and past process data, and simultaneously reduces the dimensions of time delay and the variables for feature extraction, solving the problem of the curse of dimensionality. For the dimension reduction of time delay, in order to extract similar information from the samples, each sample is reconstructed with time neighbourhoods. For the dimension reduction of the variables, considering the information of different variables variance information of the latent variables is maximized for feature extraction. Finally, a numerical example and the Tennessee Eastman process are used to demonstrate the efficacy of the proposed method.

show abstract

Batch Process Monitoring with Tensor Global–Local Structure Analysis

Cited by 35 publications

References 41 publications

A new criterion for selection of non‐zero loadings for sparse principal component analysis (SPCA)

A new criterion for selection of non‐zero loadings for sparse principal component analysis (SPCA)

Markov Chain Neighborhood Sparse Preserving Graph Embedding Based on Tensor Factorization for Batch Process Monitoring

Tensor sequence component analysis for fault detection in dynamic process

Contact Info

Product

Resources

About