Deep learning for quality prediction of nonlinear dynamic processes with variable attention‐based long short‐term memory network

Yuan, Xiaofeng; Li, Lin; Wang, Yalin; Yang, Chunhua; Gui, Weihua

doi:10.1002/cjce.23665

Cited by 83 publications

(24 citation statements)

References 44 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In recent years, deep learning methods, such as stacked auto-encoder (SAE) (Qiu and Dai, 2019; Zheng and Zhao, 2020), DBN (Yu and Liu, 2020; Yu and Yan, 2019), convolutional neural network (Zhao et al , 2019; Wu and Zhao, 2018) and long short-term memory (Han et al , 2020; Yuan et al , 2020), have been developed and widely applied in many scenarios, and they have played an important role in minimizing feature redundancy and capturing significant features. Deep learning methods are currently the most appealing feature extraction technology that has achieved brilliant success in processing high-dimensional complex data (Wang et al , 2020).…”

Section: Introductionmentioning

confidence: 99%

Industrial process data visualization based on a deep enhanced t-distributed stochastic neighbor embedding neural network

Yan

2022

View full text Add to dashboard Cite

Purpose The purpose of this paper is to propose a approach for data visualization and industrial process monitoring. Design/methodology/approach A deep enhanced t-distributed stochastic neighbor embedding (DESNE) neural network is proposed for data visualization and process monitoring. The DESNE is composed of two deep neural networks: stacked variant auto-encoder (SVAE) and a deep label-guided t-stochastic neighbor embedding (DLSNE) neural network. In the DESNE network, SVAE extracts informative features of the raw data set, and then DLSNE projects the extracted features to a two dimensional graph. Findings The proposed DESNE is verified on the Tennessee Eastman process and a real data set of blade icing of wind turbines. The results indicate that DESNE outperforms some visualization methods in process monitoring. Originality/value This paper has significant originality. A stacked variant auto-encoder is proposed for feature extraction. The stacked variant auto-encoder can improve the separation among classes. A deep label-guided t-SNE is proposed for visualization. A novel visualization-based process monitoring method is proposed.

show abstract

Section: Introductionmentioning

confidence: 99%

Industrial process data visualization based on a deep enhanced t-distributed stochastic neighbor embedding neural network

Yan

2022

View full text Add to dashboard Cite

show abstract

“…An approach to use ML to optimize makespan in job shop scheduling problems can be found in (Dao et al, 2018). Deep learning methods for example are used to predict product quality with data from parallel (Zhenyu et al, 2020) or dynamic non-linear processes (Wang & Jiao, 2017;Yuan et al, 2020). Also, there are data-driven approach for complex production systems (Ren et al, 2020).…”

Section: Introductionmentioning

confidence: 99%

Comparison of algorithms for error prediction in manufacturing with automl and a cost-based metric

et al. 2022

View full text Add to dashboard Cite

In order to manufacture products at low cost, machine learning (ML) is increasingly used in production, especially in high wage countries. Therefore, we introduce our PREFERML AutoML system, which is adapted to the production environment. The system is designed to predict production errors and to help identifying the root cause. It is particularly important to produce results for further investigations that can also be used by quality engineers. Quality engineers are not data science experts and are usually overwhelmed with the settings of an algorithm. Because of this, our system takes over this task and delivers a fully optimized ML model as a result. In this paper, we give a brief overview of what results can be achieved with a state-of-the-art classifier. Moreover, we present the results with optimized tree-based algorithms based on RandomSearchCV and HyperOpt hyperparameter tuning. The algorithms are optimized based on multiple metrics, which we will introduce in the following sections. Based on a cost-oriented metric we can show an improvement for companies to predict the outcome of later product tests. Further, we compare the results from the mentioned optimization approaches and evaluate the needed time for them.

show abstract

“…Although the LSTM based modeling issue has been studied for a long time, the history of its applications on the data-driven industry process modeling and quality prediction is only a few years. The LSTM based method has been proved as an efficient soft sensor model for nonlinear dynamic industrial processes, which can provide a better performance than the traditional models such as PLS, SAE and RNN [31].…”

Section: Introductionmentioning

confidence: 99%

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

Zheng

Shen

2021

IEEE Access

View full text Add to dashboard Cite

To predict key variables of complicated batch processes, the long short-term memory (LSTM) soft sensor is developed to deal with both data nonlinearity and dynamics. To extract proper historical samples and implement the real-time modeling scheme with model updating strategy, the just-in-time learning (JITL) algorithm is widely used at the data selection stage of LSTM soft sensor. However, the multiphase issue of batch processes are not considered for the conventional JITL-LSTM soft sensor. In this paper, a multiphase Mahalanobis distance based JITL framework is developed to integrate the phase recognition strategy into the similarity measurement and data selection scheme, by which an extra step of phase identification can be avoided and the accuracy of JITL can be significantly improved. Thus, batch samples from different operating phases can be recognized without an additional phase identification step. By the use of the Mahalanobis Distance based JITL-LSTM Soft Sensor, the probability of data mismatch can be significantly reduced so that the accuracy of quality prediction can be promoted. Two simulation cases are provided to verify the effectiveness of the proposed method consisting of a fed-batch reactor process and the penicillin fermentation process.

show abstract

Deep learning for quality prediction of nonlinear dynamic processes with variable attention‐based long short‐term memory network

Cited by 83 publications

References 44 publications

Industrial process data visualization based on a deep enhanced t-distributed stochastic neighbor embedding neural network

Industrial process data visualization based on a deep enhanced t-distributed stochastic neighbor embedding neural network

Comparison of algorithms for error prediction in manufacturing with automl and a cost-based metric

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

Contact Info

Product

Resources

About