A Framework and Modeling Method of Data-Driven Soft Sensors Based on Semisupervised Gaussian Regression

Wan, Yan; Guo, Pengju; Tian, Yu; Gao, Jianjun

doi:10.1021/acs.iecr.5b04118

Cited by 37 publications

(21 citation statements)

References 24 publications

(43 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Kadlec et al [2] have provided a detailed discussion on the design and application of data-driven soft sensors in the process industry. A few other previous works [5], [6], [20], [124]- [126] also discussed some issues/challenges relating to the design and application of soft sensors and can be found in the literature.…”

Section: A Possible Future Directions and Challengesmentioning

confidence: 99%

Design and Applications of Soft Sensors in Polymer Processing: A Review

Abeykoon

2019

IEEE Sensors J.

View full text Add to dashboard Cite

In manufacturing industry, process monitoring is a key to observe the product quality, operational health, safety and also for achieving good/satisfactory process control performance. In polymer processing, the level of control of the process operational quality is vastly dependent upon the performance of the process monitoring techniques as it is quite difficult to observe the inside of an extruder barrel (or processing chamber) during the process operation. Currently, a number of physical sensing devices/techniques are widely available for industrial applications for monitoring of parameters such as melt temperature, melt pressure, screw/processing speed, product dimensions, etc in real-time. However, there may be some limitations in using physical sensors in piratical process measurements (such as melt viscosity and melt temperature profile) due to several constraints such as their access requirements, disruptive effects to the melt flow, fragility, complexity, etc. Therefore, soft sensing techniques should be highly useful for improved process monitoring and modelling, and hence for advanced process control. In this work, a general description is presented on the state-of-the-art of soft sensor design. Then, a comprehensive review is made on the use of soft sensing techniques in polymer processing applications while identifying their capabilities and limitations. Eventually, the importance of developing of such soft sensing techniques (together with some sort of built-on intelligence) for the advancement of process monitoring is discussed while indicating some of the possible directions for future industry.

show abstract

Section: A Possible Future Directions and Challengesmentioning

confidence: 99%

Design and Applications of Soft Sensors in Polymer Processing: A Review

Abeykoon

2019

IEEE Sensors J.

View full text Add to dashboard Cite

show abstract

“…The associate editor coordinating the review of this manuscript and approving it for publication was Xiao-Sheng Si . key variables online [3]- [5]. Therefore, auxiliary variables based on limited data are more critical for modeling.…”

Section: Introductionmentioning

confidence: 99%

Modeling of Wastewater Treatment Processes Using Dynamic Bayesian Networks Based on Fuzzy PLS

et al. 2020

View full text Add to dashboard Cite

The complicated characteristics of wastewater treatment plants (WWTPs) significantly hinder the monitoring of industrial processes, and thus much attention has been paid to process modeling and prediction. A fuzzy partial least squares-based dynamic Bayesian networks (FPLS-DBN) is proposed to improve the modeling ability in WWTPs. To adapt the nonlinear process data, fuzzy partial least squares (FPLS) is introduced by using a fuzzy system to extract nonlinear features from process data. In addition, a dynamic extension is included by embedding augmented matrices into Bayesian networks to fit the uncertainty and time-varying characteristics. Regarding the quality indices for effluent suspended solid in the WWTP, the root mean square error of the FPLS-DBN model is decreased by 28.63% and 69.47%, respectively, in comparison with that for partial least squares and Bayesian networks. The results demonstrate the superiority of FPLS-DBN in modeling performance for an actual industrial WWTP application. INDEX TERMS Fuzzy partial least squares, latent variables, Bayesian networks, dynamic process modeling, wastewater treatment processes.

show abstract

“…modeling has been gained popularity in the process industries, mainly due to the fact that data-driven modeling does not need to explore the complex process mechanism exactly, but resort to the collected data [3], [4]. Among them, Principal Component Regression (PCR), Partial Least Squares (PLS) [5], Gaussian process regression (GPR) [6], Support Vector Machine (SVM) [7], Deep learning networks [8] and other models have attracted extensive attentions in industrial and academic communities.…”

Section: Introductionmentioning

confidence: 99%

“…Therefore, to develop soft sensors, traditional methods usually only use labeled data for model establishment and most unlabeled data information are fully taken for granted as useless [9]. To make full use of the information carried by unlabeled data, semi-supervised learning was proposed [3]. Semi-supervised learning method can be categorized as: graph-based method [10], generative models [11], transductive support vector machines (TSVM) [12], self-training [13] and co-training [14].…”

Section: Introductionmentioning

confidence: 99%

Learning Adaptive Semi-Supervised Multi-Output Soft-Sensors With Co-Training of Heterogeneous Models

Liu

Huang

et al. 2020

IEEE Access

View full text Add to dashboard Cite

Soft-sensors are widely utilized for predictions of important but hard-to-measure variables in industrial processes. However, significant variations, process uncertainties, negative influence of external environment and insufficient use of unlabeled data always cause the attenuation of prediction performance. Thus, this paper proposed an adaptive semi-supervised multi-output soft-sensor by co-training recursive heterogeneous models. In the proposed strategy, a linear multi-output model, called recursive partial least square (MRPLS), and a nonlinear multi-output, called long short-term memory recurrent neural network (MLSTM), are co-trained to deal with inefficient use of label data adaptively. Ensemble of both models are not only able to address the linear and nonlinear hybrid behaviors in different time scale, but also able to deal with multiple tasks learning issues. In addition, the model proposed an odd-even grouping strategy to equalize two parts of the labeled data, which is able to capture the global variations of a process. To validate the prediction performance of the proposed soft-sensor, it was verified through a simulation benchmark platform (BSM1) and a real sewage treatment plant (UCI database). The results meant that co-training MRPLS-MLSTM achieved better performance compared with other existing co-training models in terms of the hard-to-measure variables. INDEX TERMS Soft-sensors, semi-supervised, co-training, multi-output, recursive partial least square (RPLS), long short-term memory recurrent neural network (LSTM).

show abstract

A Framework and Modeling Method of Data-Driven Soft Sensors Based on Semisupervised Gaussian Regression

Cited by 37 publications

References 24 publications

Design and Applications of Soft Sensors in Polymer Processing: A Review

Design and Applications of Soft Sensors in Polymer Processing: A Review

Modeling of Wastewater Treatment Processes Using Dynamic Bayesian Networks Based on Fuzzy PLS

Learning Adaptive Semi-Supervised Multi-Output Soft-Sensors With Co-Training of Heterogeneous Models

Contact Info

Product

Resources

About