2022
DOI: 10.1002/cctc.202200355
|View full text |Cite
|
Sign up to set email alerts
|

Theory‐Guided Machine Learning to Predict the Performance of Noble Metal Catalysts in the Water‐Gas Shift Reaction

Abstract: Machine learning (ML) has widespread applications in catalyst discovery and reaction optimization. We present a theoryguided machine learning framework to evaluate the carbon monoxide (CO) conversion performance of noble metal catalysts in water-gas shift (WGS) reaction. Our study is based on an open source WGS dataset, which we modify significantly to be consistent with the chemical reaction principles. We apply state-of-the-art ML models including artificial neural networks, extreme gradient boosting to pred… Show more

Help me understand this report

Search citation statements

Order By: Relevance

Paper Sections

Select...
3
2

Citation Types

1
7
0

Year Published

2022
2022
2023
2023

Publication Types

Select...
2
2
1

Relationship

1
4

Authors

Journals

citations
Cited by 5 publications
(8 citation statements)
references
References 44 publications
1
7
0
Order By: Relevance
“…We find that the reaction temperature is the most important feature of all, where SHAP values vary from −0.6 to 0.5. This finding is consistent with the past ML studies , based on the database prepared by Odabaşı et al Moreover in Figure , each impact point of a given feature is colored by the corresponding feature value, which helps us to illustrate any generic correlation between the feature and the CO conversion. In the case of the reaction temperature, we can infer that the CO conversion and reaction temperature have a positive correlation as the colors of the impact points change from blue to red with the increase of the SHAP values.…”
Section: Resultssupporting
confidence: 91%
See 1 more Smart Citation
“…We find that the reaction temperature is the most important feature of all, where SHAP values vary from −0.6 to 0.5. This finding is consistent with the past ML studies , based on the database prepared by Odabaşı et al Moreover in Figure , each impact point of a given feature is colored by the corresponding feature value, which helps us to illustrate any generic correlation between the feature and the CO conversion. In the case of the reaction temperature, we can infer that the CO conversion and reaction temperature have a positive correlation as the colors of the impact points change from blue to red with the increase of the SHAP values.…”
Section: Resultssupporting
confidence: 91%
“…6−11 Recently, the authors of this article proposed a novel theoryguided ML model trained on the same database. 12 Theoryguided ML is an emerging field in data science where ML or deep learning models are trained with data as well as theoretical knowledge such as mathematical constraints, physics laws, and chemistry principles. 13,14 We showed that the theory-guided ML model predicts CO conversion that strictly obeys the thermodynamic equilibrium principle.…”
Section: Introductionmentioning
confidence: 99%
“…The calculated RF Importance of the descriptors with CO conversion as the target variable is reported on Figure , along with this cumulative parameter of the different descriptors’ subcategories in the data set. As reported previously by the works of Suzuki and Chattoraj, , the descriptor that holds the greatest calculated Importance is the reaction temperature, and the descriptors associated with the experimental conditions hold, cumulatively, the vast majority of the Importance, at around 70%, with the rest of the variables (catalyst components, preparation methods, and composition) comprising the rest.…”
Section: Resultsmentioning
confidence: 99%
“…The developed and reported a model based on Least Squares Support-Vector Machine (LSSVM) coupled with Particle Swarm Optimization to predict catalytic H 2 generation, but focused more on model optimization . More recently, Chattoraj et al have reported a theory-guided ML framework to enhance the predictive capacity by coupling an ANN module with a custom loss function that takes into account the reaction’s thermodynamics, and predicting CO conversion with an Extreme Gradient Boosting model, while employing Odabaşi’s data set after making modifications to correct inconsistencies in a substantial portion of the entries …”
Section: Introductionmentioning
confidence: 99%
See 1 more Smart Citation