Capabilities of Machine Learning Methods in Prediction of Solubility of Substances in Supercritical Carbon Dioxide

Abstract

This work reviews studies of the application of machine learning methods and neural network technologies in the prediction of solubility of various substances in supercritical fluids. By the example of an existing data set on the solubility of aromatic hydrocarbons in supercritical carbon dioxide, a prototype of a solubility prediction system was developed using a simple three-layer neural network. Its efficiency is shown, and further directions of research in this field are identified.