Machine learning based interpretation of microkinetic data: a Fischer–Tropsch synthesis case study

Abstract

Machine-Learning (ML) methods, such as Artificial Neural Networks (ANN) bring the data-driven design of chemical reactions within reach. Simultaneously with the verification of the absence of any bias in the...

Quo vadis multiscale modeling in reaction engineering? – A perspective

“…Most important is to retain the accuracy on the model simulations, as e.g., successfully demonstrated for the Fischer-Tropsch synthesis using a ML model trained using data generated from a microkinetic model. 172 Interpretability techniques have been used to demonstrate that the trends predicted as a function of the operating variables is indeed present in the ML model, albeit that no fundamental relationship can be formulated. Nevertheless, as ML model simulated outlet values closely correspond to the ones simulated with detailed models, ML models can be envisaged as a potential tool for upscaling purposes.…”

Micromixing Performance and Residence Time Distribution in a Miniaturized Magnetic Reactor: Experimental Investigation and Machine Learning Modeling

“…A larger Shapley value represents larger marginal contribution. The principles and procedures of the SHAP method have been illustrated by Chakkingal et al in detail …”

“…The principles and procedures of the SHAP method have been illustrated by Chakkingal et al in detail. 60…”

Multi-output machine learning models for kinetic data evaluation : A Fischer–Tropsch synthesis case study

“…Studies using machine learning methods such as neural network (NN) [39,40] and support vector regression (SVR) [41] to predict a single response FTS process have also been reported in literature, as an alternative to detailed kinetic models. Compared to that, only a limited number publications on ML methods for a multi-response scenario [42,43], such as conversion and selectivities in the FTS process, have been reported.…”