Evaluation of state-specific transport properties using machine learning methods

In this study, machine learning algorithms are employed to calculate state-to-state transport coefficients in nonequilibrium reacting gas flows. The focus is on the evaluation of thermal conductivity, shear viscosity, and bulk viscosity under conditions strong coupling between vibrational-chemical kinetics dynamics. order solve a regression problem for evaluating coefficients, specific software application with user interface developed, which allows loading, processing, saving data arrays; configuring model architecture; training models various optimizers, loss functions, metrics; making predictions using trained models. Using developed multi-layer perceptron constructed trained. assessed binary mixture molecular atomic nitrogen taking into account 48 vibrational states; computed wide temperature range varying composition. Good agreement results original calculated rigorous but computationally expensive kinetic theory shown. Applying techniques yields significant speedup about two orders magnitude computation coefficients. It concluded that implementation methods may considerably reduce computational efforts required flow simulations.