Meta-modelling in chemical process system engineering

Kajero, Olumayowa T.; Chen, Tao; Yao, Yuan; Chuang, Yao-Chen; Wong, David Shan Hill

doi:10.1016/j.jtice.2016.10.042

Cited by 27 publications

(16 citation statements)

References 189 publications

(125 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…heat transfer coefficients) and, possibly, different units/reactors geometries, etc. [6,7]. Also, since process FPMs typically do not take into account the physical characteristics of mechanical and electrical components, connections and piping, which remarkably influence the real process, the accuracy of the FPMs predictions are reduced [8,9].…”

Section: ) Three-tanks Systemmentioning

confidence: 99%

“…The most common techniques include Latin hypercube sampling [44], low discrepancy sequences as the Hammersley technique [58] and space-filling designs like max-min techniques and space-filling Latin hypercube design [59]. Alternatively, sequential or adaptive sampling are special type of DOCE techniques that are typically related to the use of OK/GP models [7]. In these sequential techniques, the total number of training points are not selected at once: the surrogate model is initially fitted with a relatively small number of training points, and it is iteratively updated by adding new training points of interest (infill or update points) to the initial training dataset so as to enhance a desired criterion of the metamodel performance, which is highly dependent on the eventual use of the metamodel (global approximation, surrogate-based optimization, etc.…”

Section: Design Of Computer Experiments (Doce)mentioning

confidence: 99%

“…Specifically, for any of the models in Eqs. (7), four different ANNs-based dynamic models, corresponding to four different lags ( = 0, 1, 2 3), have been fitted. Since in each case the number of the model inputs will be different, a single fixed ANN structure is not likely to be suitable for all these different dynamic models.…”

Section: Figure 3 Training Signal (Bioreactor)mentioning

confidence: 99%

See 2 more Smart Citations

Dynamic Surrogate Modeling for Multistep-ahead Prediction of Multivariate Nonlinear Chemical Processes

Shokry

Baraldi

Zio

et al. 2020

Ind. Eng. Chem. Res.

View full text Add to dashboard Cite

This work proposes a methodology for multivariate dynamic modeling and multistep-ahead prediction of nonlinear systems using surrogate models for the application to nonlinear chemical processes. The methodology provides a systematic and robust procedure for the development of data-driven dynamic models capable of predicting the process outputs over long time horizons. It is based on using surrogate models to construct several Nonlinear AutoRegressive eXogenous models (NARX), each one approximating the future behavior of one process output as a function of the current and previous process inputs and outputs. The developed dynamic models are employed in a recursive schema to predict the process future outputs over several time steps (multistep-ahead prediction). The methodology is able to manage two different scenarios: 1) one in which a set of input-output signals collected from the process is only available for training, and 2) another in which a mathematical model of the process is available and can be used to generate specific datasets for training. With respect to the latter, the proposed methodology includes a specific procedure for the selection of training data in dynamic modeling based on Design Of Computer Experiment (DOCE) techniques. The proposed methodology is applied to case studies from the process industry presented in the literature. The results show very high prediction accuracies over long time horizons. Also, thanks to the flexibility, robustness and computational efficiency of surrogate modeling, the methodology allows dealing with a wide range of situations, which would be difficult to address using first principle models.

show abstract

Section: ) Three-tanks Systemmentioning

confidence: 99%

Section: Design Of Computer Experiments (Doce)mentioning

confidence: 99%

Section: Figure 3 Training Signal (Bioreactor)mentioning

confidence: 99%

See 1 more Smart Citation

Dynamic Surrogate Modeling for Multistep-ahead Prediction of Multivariate Nonlinear Chemical Processes

Shokry

Baraldi

Zio

et al. 2020

Ind. Eng. Chem. Res.

View full text Add to dashboard Cite

show abstract

“…Alternatively, computational fluid dynamics (CFD) models utilized in multiphase flows can provide useful information [5]- [8], [11], [12], [17], [18]. A few studies investigated the temperature characteristics of twin-screw multiphase pumps [12].…”

Section: A Motivationmentioning

confidence: 99%

Special Probabilistic Prediction Model for Temperature Characteristics of Dynamic Fluid Processes

et al. 2019

View full text Add to dashboard Cite

Accurately predicting the temperature characteristics of a dynamic discharge process in different transportation conditions can improve the performance of reciprocating multiphase pumps in practice. However, an accurate model for the description of the complicated behavior is not available because of the unknown interphase interaction mechanisms and infeasible experiments. A probabilistic modeling method of automatically selecting prediction models is proposed for the dynamic discharge process. First, candidate computational fluid dynamics (CFD) models are empirically utilized to provide the training data for candidate Gaussian process models (GPMs). Then, a posterior probability index is proposed to assess the uncertainty of trained GPMs when the actual values are not available. With this information, the most suitable GPM and CFD models are selected sequentially for each new sample. Consequently, the developed special GPM (SGPM) can capture the main temperature characteristics. Moreover, the selection results of prediction models can provide useful information for the recognition of complicated flow patterns. The advantages of the proposed SGPM are demonstrated using a reciprocating multiphase pump under different transportation conditions.INDEX TERMS Probabilistic modeling, gaussian process model, computational fluid dynamics, multiphase pump.

show abstract

“…135-178) and Woods and Lewis (2016) with its 117 references. Alternative methods for reducing the dimensionality use "moving least squares" (MLS), "partial least squares" (PLS), and "principal component analysis" (PCA): see Wang et al (2011), Bouhlel et al ( , 2017, and Kajero et al (2017), respectively.…”

Section: Introductionmentioning

confidence: 99%

Prediction for Big Data Through Kriging: Small Sequential and One-Shot Designs

Kleijnen

Beers

2018

SSRN Journal

View full text Add to dashboard Cite

Kriging or Gaussian process (GP) modeling is an interpolation method that assumes the outputs (responses) are more correlated, the closer the inputs (explanatory or independent variables) are. A GP has unknown (hyper)parameters that must be estimated; the standard estimation method uses the "maximum likelihood" criterion. However, big data make it hard to compute the estimates of these GP parameters, and the resulting Kriging predictor and the variance of this predictor. To solve this problem, some authors select a relatively small subset from the big set of previously observed "old" data; their method is sequential and depends on the variance of the Kriging predictor. The resulting designs turn out to be "local"; i.e., most design points are concentrated around the point to be predicted. We develop three alternative one-shot methods that do not depend on GP parameters: (i) select a small subset such that this subset still covers the original input space-albeit coarser; (ii) select a subset with relatively many-but not all-combinations close to the new combination that is to be predicted, and (iii) select a subset with the nearest neighbors (NNs) of this new combination. To evaluate these designs, we compare their squared prediction errors in several numerical (Monte Carlo) experiments. These experiments show that our NN design is a viable alternative for the more sophisticated sequential designs.

show abstract

Meta-modelling in chemical process system engineering

Cited by 27 publications

References 189 publications

Dynamic Surrogate Modeling for Multistep-ahead Prediction of Multivariate Nonlinear Chemical Processes

Dynamic Surrogate Modeling for Multistep-ahead Prediction of Multivariate Nonlinear Chemical Processes

Special Probabilistic Prediction Model for Temperature Characteristics of Dynamic Fluid Processes

Prediction for Big Data Through Kriging: Small Sequential and One-Shot Designs

Contact Info

Product

Resources

About