CALCULATION OF THE PROPERTIES OF HYDROCARBON MIXTURES USING ARTIFICIAL NEURAL NETWORKS

This article discusses replacing the classical approaches to calculating the properties of hydrocarbons and their mixtures with an artificial neural network. In the work, a data sample was created, representing more than 18,000 variants of model mixtures. Boiling points, molecular weights, and densities were calculated for each mixture. On the basis of the received training sample, a neural network was created, which allows calculating the properties based on a limited set of initial data: mass concentrations of its components and their normal boiling points