Bayesian chemical reaction neural network for autonomous kinetic uncertainty quantification

Li, Qiaofeng; Chen, Huaibo; Koenig, Benjamin C.; Deng, Sili

doi:10.1039/d2cp05083h

Cited by 10 publications

(9 citation statements)

References 44 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…It is the goal of this work to extend the CRNN with a Bayesian framework to increase its utility in QSP and other scientific domains. This is similar to the work by Li et al in which they also extend the CRNN to include a Bayesian framework [5], but here we use a different methodology for sampling the posterior.…”

Section: Introductionmentioning

confidence: 78%

Uncertainty quantified discovery of chemical reaction systems via Bayesian scientific machine learning

Nieves,

Dandekar,

Rackauckas

2023

Preprint

View full text Add to dashboard Cite

The recently proposed Chemical Reaction Neural Network (CRNN) discovers chemical reaction pathways from time resolved species concentration data in a deterministic manner. Since the weights and biases of a CRNN are physically interpretable, the CRNN acts as a digital twin of a classical chemical reaction network. In this study, we employ a Bayesian inference analysis coupled with neural ordinary differential equations (ODEs) on this digital twin to discover chemical reaction pathways in a probabilistic manner. This allows for estimation of the uncertainty surrounding the learned reaction network. To achieve this, we propose an algorithm which combines neural ODEs with a preconditioned stochastic gradient langevin descent (pSGLD) Bayesian framework, and ultimately performs posterior sampling on the neural network weights. We demonstrate the successful implementation of this algorithm on several reaction systems by not only recovering the chemical reaction pathways but also estimating the uncertainty in our predictions. We compare the results of the pSGLD with that of the standard SGLD and show that this optimizer more efficiently and accurately estimates the posterior of the reaction network parameters. Additionally, we demonstrate how the embedding of scientific knowledge improves extrapolation accuracy by comparing results to purely data-driven machine learning methods. Together, this provides a new framework for robust, autonomous Bayesian inference on unknown or complex chemical and biological reaction systems.

show abstract

Section: Introductionmentioning

confidence: 78%

Uncertainty quantified discovery of chemical reaction systems via Bayesian scientific machine learning

Nieves,

Dandekar,

Rackauckas

2023

Preprint

View full text Add to dashboard Cite

show abstract

“…Although PINNs and further kinetic ML approaches were previously introduced, − ,,− it should be noted that the influence of external parameter variation was not taken into detailed account. However, reaction optimization requires a sufficient sampling of the parameter space, such that the benefits of the previous models for process design concepts are limited .…”

Section: Discussionmentioning

confidence: 99%

“…In terms of such probabilistic approaches, the resulting calculations also provide reasonable estimates for the uncertainty of the kinetic outcomes. 59 PINNs were also used for the detection of unknown chemical pathways and the inference of rate constants from experimental data. It has to be noted that the corresponding approaches mainly focus on activation energies in combination with standard Arrhenius kinetics.…”

Section: Introductionmentioning

confidence: 99%

“…It has to be noted that the corresponding approaches mainly focus on activation energies in combination with standard Arrhenius kinetics . Despite recent developments, ,− a detailed study of external parameter variations such as temperature and start concentration changes and their influence on the resulting concentration profiles was out of scope.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Scientific Deep Machine Learning Concepts for the Prediction of Concentration Profiles and Chemical Reaction Kinetics: Consideration of Reaction Conditions

Adebar,

Keupp,

Emenike

et al. 2024

J. Phys. Chem. A

View full text Add to dashboard Cite

Emerging concepts from scientific deep machine learning such as physics-informed neural networks (PINNs) enable a data-driven approach for the study of complex kinetic problems. We present an extended framework that combines the advantages of PINNs with the detailed consideration of experimental parameter variations for the simulation and prediction of chemical reaction kinetics. The approach is based on truncated Taylor series expansions for the underlying fundamental equations, whereby the external variations can be interpreted as perturbations of the kinetic parameters. Accordingly, our method allows for an efficient consideration of experimental parameter settings and their influence on the concentration profiles and reaction kinetics. A particular advantage of our approach, in addition to the consideration of univariate and multivariate parameter variations, is the robust model-based exploration of the parameter space to determine optimal reaction conditions in combination with advanced reaction insights. The benefits of this concept are demonstrated for higher-order chemical reactions including catalytic and oscillatory systems in combination with small amounts of training data. All predicted values show a high level of accuracy, demonstrating the broad applicability and flexibility of our approach.

show abstract

“…Uncertainty quantification is an important aspect of modeling a chemical reaction, because it ensures that the model is reliable and accurate in further applications. Li et al developed Bayesian CRNN (B-CRNN) based on the deterministic CRNN and Bayesian inference, where the physical parameters (network weights) are given probabilistic distributions instead of being set as deterministic. The case studies demonstrated that the B-CRNN can simultaneously infer the chemical reaction pathways from measured noisy concentration data and perform uncertainty quantification of the identified reaction pathways, kinetic parameters, and concentration predictions with unseen initial conditions.…”

Section: The Application In Modeling Chemical Processesmentioning

confidence: 99%

The Application of Physics-Informed Machine Learning in Multiphysics Modeling in Chemical Engineering

Wu,

Wang,

et al. 2023

Ind. Eng. Chem. Res.

View full text Add to dashboard Cite

Physics-Informed Machine Learning (PIML) is an emerging computing paradigm that offers a new approach to tackle multiphysics modeling problems prevalent in the field of chemical engineering. These problems often involve complex transport processes, nonlinear reaction kinetics, and multiphysics coupling. This Review provides a detailed account of the main contributions of PIML with a specific emphasis on modeling momentum transfer, heat transfer, mass transfer, and chemical reactions. The progress in method development (e.g., algorithm and architecture), software libraries, and specific applications (e.g., multiphysics coupling and surrogate modeling) are detailed. On this basis, future challenges highlight the importance of developing more practical solutions and strategies for PIML, including turbulence models, domain decomposition, training acceleration, surrogate modeling, hybrid modeling, and geometry module creation.

show abstract

Bayesian chemical reaction neural network for autonomous kinetic uncertainty quantification

Abstract: Chemical reaction neural network (CRNN), a recently developed tool for autonomous discovery of reaction models, has been successfully demonstrated on a variety of chemical engineering and biochemical systems. It leverages...

Cited by 10 publications

References 44 publications

Uncertainty quantified discovery of chemical reaction systems via Bayesian scientific machine learning

Uncertainty quantified discovery of chemical reaction systems via Bayesian scientific machine learning

Scientific Deep Machine Learning Concepts for the Prediction of Concentration Profiles and Chemical Reaction Kinetics: Consideration of Reaction Conditions

The Application of Physics-Informed Machine Learning in Multiphysics Modeling in Chemical Engineering

Contact Info

Product

Resources

About