Profitability related industrial-scale batch processes monitoring via deep learning based soft sensor development

Ji, C.; Ma, Fangyuan; Sun, Wei

doi:10.1016/j.compchemeng.2022.108125

Cited by 27 publications

(11 citation statements)

References 38 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Batch process is characterized by the batch characteristic, and therefore, the production data have a three‐dimensional characteristic. Three‐dimensional data

X ((), I \times J \times K)

(where

I

denotes the batch,

J

denotes the variable, and

K

denotes the sample) cannot be used directly in quality prediction, and therefore, data preprocessing is required 34 . Assuming that the data for a particular batch of batch process is

\overset{X}{true} \in R^{k \times j}

, then the data can be expressed:

\overset{X}{true} goodbreak= ([],,,, x_{1}, x_{2}, \dots, x_{j}) goodbreak= ([], \begin{array}{c} x_{11} & x_{21} & \dots & x_{j 1} \\ x_{12} & x_{22} & \dots & x_{j 2} \\ ⋮ & ⋮ & \dots & ⋮ \\ x_{1 k} & x_{2 k} & \dots & x_{jk} \end{array})

where

k

denotes the number of samples in the batch and

j

denotes the number of the variables.…”

Section: The Proposed Dsca‐tcn Quality Monitoring Model For Batch Pro...mentioning

confidence: 99%

“…In the process of model training, the value of batch size is positively correlated with the accuracy of gradient descent. The batch size here is not related to the batch process but is a hyperparameter, that is, the hyperparameter equals the amount of data that is used in each network training in the deep learning model 34 . However, the batch size is limited by the capacity of the GPU, and a large batch size increases the training time of the model.…”

Section: Case Studiesmentioning

confidence: 99%

“…Three-dimensional data X I Â J Â K ð Þ(where I denotes the batch, J denotes the variable, and K denotes the sample) cannot be used directly in quality prediction, and therefore, data preprocessing is required. 34 Assuming that the data for a particular batch of batch process is X R kÂj , then the data can be expressed:…”

Section: Data Unfoldingmentioning

confidence: 99%

“…The batch size here is not related to the batch process but is a hyperparameter, that is, the hyperparameter equals the amount of data that is used in each network training in the deep learning model. 34 However, the batch size is limited by the capacity of the GPU, and a large batch size increases the training time of the model. The number of iterations and batch size of all comparison models are set to 60 and 70, respectively.…”

Section: The Selection Of Model Parametersmentioning

confidence: 99%

See 3 more Smart Citations

Batch process quality monitoring based on temporal convolutional networks with depthwise separable coordinated attention module

Zhao,

Tuo,

Mou

et al. 2023

Asia-Pacific J Chem Eng

View full text Add to dashboard Cite

Quality monitoring is an important tool for ensuring the safe operation of batch processes and the high quality of the final products. However, the inherent non‐linear, dynamic, and batch characteristics make the quality monitoring of batch process still have some difficulties. To solve these problems, this paper proposes a batch quality monitoring model based on a temporal convolutional network with a depthwise separable coordinated attention module. Firstly, a method of data unfolding incorporating sliding windows is proposed to unfold and stack the data along the direction of the variables, and a variable selection method of maximum information coefficient fused with Monte Carlo sampling is proposed to select the process variables related to the quality variables. Secondly, we take the traditional temporal convolutional network as the base network and decouple the correlation between the batch data by using depthwise separable convolution. At the same time, we utilize coordinate attention to extract data features in different spatial directions to ensure the effectiveness of quality monitoring. Finally, the feasibility and robustness of the proposed model are verified by a nonlinear numerical example and an industrial‐scale penicillin fermentation process. The experimental results show the proposed model has lower false alarm rate and false negative rate and can be used to maintain the product quality of the actual batch process.

show abstract

“…Batch process is characterized by the batch characteristic, and therefore, the production data have a three‐dimensional characteristic. Three‐dimensional data

X ((), I \times J \times K)

(where

I

denotes the batch,

J

denotes the variable, and

K

denotes the sample) cannot be used directly in quality prediction, and therefore, data preprocessing is required 34 . Assuming that the data for a particular batch of batch process is

\overset{X}{true} \in R^{k \times j}

, then the data can be expressed:

\overset{X}{true} goodbreak= ([],,,, x_{1}, x_{2}, \dots, x_{j}) goodbreak= ([], \begin{array}{c} x_{11} & x_{21} & \dots & x_{j 1} \\ x_{12} & x_{22} & \dots & x_{j 2} \\ ⋮ & ⋮ & \dots & ⋮ \\ x_{1 k} & x_{2 k} & \dots & x_{jk} \end{array})

where

k

denotes the number of samples in the batch and

j

denotes the number of the variables.…”

Section: The Proposed Dsca‐tcn Quality Monitoring Model For Batch Pro...mentioning

confidence: 99%

Section: Case Studiesmentioning

confidence: 99%

Section: Data Unfoldingmentioning

confidence: 99%

Section: The Selection Of Model Parametersmentioning

confidence: 99%

See 2 more Smart Citations

Batch process quality monitoring based on temporal convolutional networks with depthwise separable coordinated attention module

Zhao,

Tuo,

Mou

et al. 2023

Asia-Pacific J Chem Eng

View full text Add to dashboard Cite

show abstract

“…Data-driven quality prediction models usually transform process variables into potential feature space and mine the feature information of data to achieve the expected quality indicator variables (QIVs). Generally speaking, two kinds of methods can be used to mine characteristic information in historical data, namely the machine learning method based on multivariate statistical analysis and the deep learning method [12][13][14]. At present, some researchers have successfully applied multivariate statistical machine learning methods to quality prediction, such as principal component regression (PCR) [15,16], partial least squares (PLS) [17,18], support vector regression [19,20] and neighborhood preserving embedding [21,22].…”

Section: Introductionmentioning

confidence: 99%

Deep learning with CBAM-based CNN for batch process quality prediction

Zhao,

Tuo,

Hui

2023

Meas. Sci. Technol.

View full text Add to dashboard Cite

Data-driven quality prediction method has been widely used in product estimation of batch processes. However, the initial conditions of different batches in batch process are different, and the multiphase characteristics and nonlinearity in batch are not conducive to the quality prediction. To solve these problems, a model for batch process quality prediction based on a convolutional neural network (CNN) is proposed. Firstly, in order to enhance data characteristics and reduce model computing time, a maximum information coefficient (MIC) method based on mutual information is used to select variables according to the correlation between process variables and quality variables. Secondly, the quality prediction model of CBAM-CNN based on the attention mechanism is established. On the one hand, an improved convolutional block attention module (CBAM) is added to the CNN. The input feature mapping is re-calibrated to focus on useful feature information and weaken irrelevant redundant information in each sliding window. On the other hand, by introducing an improved convolutional module with double-band skip connection lines, the backpropagation speed of the CBAM-CNN model is increased, which can effectively avoid the occurrence of the overfitting problem. Finally, the data of batch process is used as the input of the prediction model. The superiority and effectiveness of the proposed model are verified by predicting the quality variable of the penicillin fermentation process simulation benchmark and the industrial-scale penicillin fermentation process. It is proved that the proposed model has better generalization performance in the quality prediction of the penicillin fermentation process with different control strategies.

show abstract

Dynamic nonlinear soft sensor modelling method using linear slow feature analysis and least squares support vector regression for batch processes

Zhang,

Liu,

Zhang

et al. 2023

Can J Chem Eng

View full text Add to dashboard Cite

Data‐driven soft sensor models have been extensively utilized in industrial processes. Batch processes are usually employed to manufacture low‐volume and high value‐added products in chemical, materials, and pharmaceutical industries. The most distinctive features of batch process lie in nonlinear, repetition, and slow time varying characteristics. In this paper, a data‐driven soft sensor modelling method based on linear slow feature analysis (LSFA) and least squares support vector regression (LSSVR) is proposed. In this method, LSFA was used to effectively capture the driving force behind the data features that change as slowly as possible. Then, a LSSVR model was constructed between the extracted slow feature variables and quality variables. Finally, a numerical example, industrial penicillin fermentation processes, and cobalt oxalate synthesis process were utilized to confirm the prediction accuracy and model reliability of the proposed approach.

show abstract

Profitability related industrial-scale batch processes monitoring via deep learning based soft sensor development

Cited by 27 publications

References 38 publications

Batch process quality monitoring based on temporal convolutional networks with depthwise separable coordinated attention module

Batch process quality monitoring based on temporal convolutional networks with depthwise separable coordinated attention module

Deep learning with CBAM-based CNN for batch process quality prediction

Dynamic nonlinear soft sensor modelling method using linear slow feature analysis and least squares support vector regression for batch processes

Contact Info

Product

Resources

About