Improved Mahalanobis Distance Based JITL-LSTM Soft Sensor for Multiphase Batch Processes

Zheng, Jiaqi; Shen, Feifan; Ye, Lingjian

doi:10.1109/access.2021.3079184

Cited by 14 publications

(8 citation statements)

References 40 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Zheng et al proposed a JITL framework based on multiphase Markov distance and combined it with the long short term memory (LSTM) algorithm to update the model parameters in real-time. [15] This method improved the prediction accuracy of JITL to a larger extent and was validated in two simulation examples. In order to solve the problem of the scarcity of data on quality variables in industrial processes, Yuan et al proposed a novel semisupervised JITL framework to estimate the output by constructing a weighted probabilistic principal component regression model.…”

Section: Introductionmentioning

confidence: 94%

Online prediction based on the Copula function for the intrinsic viscosity of polyester fibre

et al. 2022

View full text Add to dashboard Cite

The polymerization process, as the starting point of polyester fibre production, directly affects the fibre properties. However, the polymerization process has complex reaction mechanisms and involves many variables, making online modelling more challenging. In this paper, we introduce the Copula function into the modelling of the polymerization process. First, a feature selection method based on kernel principal component analysis (KPCA)-Copula is proposed, and then, the Copula function to obtain the nonlinear characteristics among the variables is used. On this basis, an online prediction model of intrinsic viscosity based on just-in-time learning (JITL) was established, that is, empirical Copula-JITL. In order to improve the computational efficiency of the empirical Copula-JITL, the t-Copula function is introduced. Furthermore, by combining the t-Copula function with the empirical Copula function, we get a JITL model based on the hybrid Copula function, that is, hybrid Copula-JITL. Finally, actual polymerization process data collected from fibre production are provided to demonstrate the effectiveness of the proposed algorithms.

show abstract

Section: Introductionmentioning

confidence: 94%

Online prediction based on the Copula function for the intrinsic viscosity of polyester fibre

et al. 2022

View full text Add to dashboard Cite

show abstract

“…It mainly As can be seen from Figure 1, the definition of a similarity measure is the key to constructing high-performance JIT soft sensor models. The most commonly used similarity criteria include distance-based similarity functions, such as Euclidean distance [34,35] and Mahalanobis distance [36], angle-based similarity functions, such as cosine similarity [37,38], and similarity functions based on correlation coefficients, such as Pearson correlation coefficient similarity [39]. Among them, Euclidean-distance similarity is the simplest and most commonly used similarity criterion for JIT soft sensor modeling.…”

Section: Stacked Autoencodermentioning

confidence: 99%

Deep Semi-Supervised Just-in-Time Learning Based Soft Sensor for Mooney Viscosity Estimation in Industrial Rubber Mixing Process

et al. 2022

View full text Add to dashboard Cite

Soft sensor technology has become an effective tool to enable real-time estimations of key quality variables in industrial rubber-mixing processes, which facilitates efficient monitoring and a control of rubber manufacturing. However, it remains a challenging issue to develop high-performance soft sensors due to improper feature selection/extraction and insufficiency of labeled data. Thus, a deep semi-supervised just-in-time learning-based Gaussian process regression (DSSJITGPR) is developed for Mooney viscosity estimation. It integrates just-in-time learning, semi-supervised learning, and deep learning into a unified modeling framework. In the offline stage, the latent feature information behind the historical process data is extracted through a stacked autoencoder. Then, an evolutionary pseudo-labeling estimation approach is applied to extend the labeled modeling database, where high-confidence pseudo-labeled data are obtained by solving an explicit pseudo-labeling optimization problem. In the online stage, when the query sample arrives, a semi-supervised JITGPR model is built from the enlarged modeling database to achieve Mooney viscosity estimation. Compared with traditional Mooney-viscosity soft sensor methods, DSSJITGPR shows significant advantages in extracting latent features and handling label scarcity, thus delivering superior prediction performance. The effectiveness and superiority of DSSJITGPR has been verified through the Mooney viscosity prediction results from an industrial rubber-mixing process.

show abstract

“…However, the Euclidean distance does not consider the possible coupling relationship among features and cannot effectively describe the global relationship between the two sample points, which will often fail to achieve satisfactory results. Compared with Euclidean distance, Mahalanobis distance [17][18][19] overcomes the above shortcomings. For one thing, Mahalanobis distance does not depend on the scale, which calculates the true distance between samples under consideration of the distribution of the whole data set.…”

Section: Introductionmentioning

confidence: 99%

A multi‐fault diagnosis method based on improved SMOTE for class‐imbalanced data

Zhao

et al. 2022

Can J Chem Eng

View full text Add to dashboard Cite

With the development of industrial processes, how to effectively diagnose the faults in an increasingly complex production process has attracted widespread attention. It is worth noting that there may be multiple types of faults in the actual industrial process, and there is an extreme class imbalance between the normal samples and the fault samples. Therefore, it is of practical significance to carry out research on the multi-fault diagnosis method for class-imbalanced data. In this paper, a multi-fault diagnosis method based on improved synthetic minority sampling technology (SMOTE) is proposed. First, aiming at the class imbalance, an improved SMOTE algorithm based on Mahalanobis distance (Mahalanobis distance-based SMOTE [MSMOTE]) is proposed for over-

show abstract

Improved Mahalanobis Distance Based JITL-LSTM Soft Sensor for Multiphase Batch Processes

Cited by 14 publications

References 40 publications

Online prediction based on the Copula function for the intrinsic viscosity of polyester fibre

Online prediction based on the Copula function for the intrinsic viscosity of polyester fibre

Deep Semi-Supervised Just-in-Time Learning Based Soft Sensor for Mooney Viscosity Estimation in Industrial Rubber Mixing Process

A multi‐fault diagnosis method based on improved SMOTE for class‐imbalanced data

Contact Info

Product

Resources

About