Binary Diffusion Coefficients of Low-Density Gases by neural network

A feed-forward multi-layer neural network with Levenberg–Marquardt training algorithm was developed to predict yield for binary diffusivity coefficients of Low-Density gases.Diffusivity coefficient depends on nine independent variables including , critical temperature, critical pressure, acentric factor and molecular weight of each component and system temperature. These nine inputs were devoted to the network. Comparing predicted results of the neural network model and the mathematical model to experimental data indicated that the neural network model had better predictability than the mathematical model.

Keywords:

Neural network , Binary diffusion coefficients , low density gases