Abstract
AbstractWe propose a new approach for sampling domain reduction for efficient surrogate model generation. Currently, the standard procedure is to use box constraints for the independent variables when sampling the exact simulator. However, by including additional inequality constraints to account for interdependencies between these variables, we can drastically reduce the sampling domain and ensure consistency of unit operations. Moreover, we present a methodology for constructing surrogate models based on penalized regression and error‐maximization sampling. All these algorithms have been implemented as a free and open‐source software package. Through a case study on the water–gas shift reaction for hydrogen production, we show that sampling domain reduction reduces the required number of sampling points significantly and improves the accuracy of the surrogate model.
Highlights
Surrogate models, known as reduced-order models, response-surface models, metamodels, or proxy models, are a class of regression models that have gained much attention in recent years
We present and demonstrate the merits of our proposed approach on an auto-thermal reformer (ATR) and the water–gas shift section of hydrogen production with two reactors
We have in this article developed a methodology for constructing surrogate models based on penalized regression, error-maximization sampling, and sampling-domain reduction
Summary
Known as reduced-order models, response-surface models, metamodels, or proxy models, are a class of regression models that have gained much attention in recent years. A variation of Æ20% in the molar ratio of steam and methane can be considered sufficient for the subsequent applications of a surrogate model If we incorporate such proportional dependencies, we can reduce the sampling domain to about 40% of the domain required with box constraints. In [5], n_ di at ∈ Rndat corresponds to the inlet flow rate of chemical component i sampled from the outlet of the last subsection, ndat to the number of sampled data from the previous section for creating the constraints, and ni ∈ R to the inlet flow rate of chemical component i as used as an independent variable to the surrogate model. We created a validation set fxkg by sampling N 1⁄4 2,000 new points within the box constraint sampling domain (reforming section) and the constrained region (water–gas shift section) to evaluate the performance of the generated surrogate models. Problems when the constraints do not represent the overall system sufficiently well
Sign-in/Register to view highlights & summary 
Published Version (
Free)
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call