Artificial neural network model for the evaluation of chemical kinetics in thermally induced solid-state reaction

A regression model for plasma reaction kinetics

“…Kuang and Xu [67] showcased the use of a convolutional neural network for the prediction of kinetic triplets for pyrolysis processes from experimental data, more specifically the temperatures at preselected values of conversion degrees. Very similar work has also been done by Huang et al [68], and Vieira and Krems [69]. In most cases, a research work intersecting ML and chemical kinetics introduces ANNs and other ML model techniques (or soft computing) as an alternative to the hard kinetic model of a system, which typically integrates the differential equations governing the species densities to calculate the reaction rates.…”

Artificial Neural Networks for Pyrolysis, Thermal Analysis, and Thermokinetic Studies: The Status Quo

“…On the other hand, evaluation of the kinetic triplets with the aid of ANN is possible without explicitly separating the and terms as well as making any assumptions about their mathematical form. Some evaluations of this type [ 21 , 94 , 95 ] have already been thoroughly tested in terms of predictions outside of the data space where the ANN is trained. However, such evaluations are based on the neural networks trained on the data simulated according to Equation (1), so that the knowledge of Equation (1) implicitly propagates into predictions.…”

“…An idea of using the neural networks primarily for determination of the kinetic triplets has been put forward by Sbirrazzuoli et al [ 9 ] and further developed by Conesa et al [ 56 ], Muravyev and Pivkina [ 21 ], and other more recent workers [ 61 , 94 , 95 , 96 , 97 ]. As the approach develops, the complexity of the reaction domain has been increasing.…”

“…Huang et al [ 94 ] have adopted the same approach, but further increased the number of reaction types to 17, and utilized a more advanced mathematical tool, the general regression neural network. Most importantly, the authors employ datasets of simulated nonisothermal experiments at several different heating rates.…”

“…Kuang and Xu [ 95 ] have implemented the most advanced mathematical approach to this problem, i.e., the one-dimensional convolutional neural network (CNN). As in the previously considered work [ 94 ], the single neural network was utilized to estimate the whole kinetic triplet. The application of the trained CNN model for estimating the preexponential factor and activation energy has resulted in less than 3% error in case of using a dataset of three heating rates for training.…”