Neural Network Classification of Phase Equilibrium Methods

S. Oreški, J. Zupan and P. Glavič

Abstract
In the paper Kohonen neural network is described as an alternative tool for a fast selection of the most suitable physical property estimation method to be used in efficient chemical process design and simulation. Kohonen neural networks are trained to suggest the appropriate method of phase equilibrium estimation on the basis of known physical properties of samples (objects of the study). In other words, they classify the objects into none, one or more possible classes (possible methods of phase equilibrium) and estimate the reliability of the proposed classes (adequacy of different methods of phase equilibrium). Kohonen map with almost clearly separated clusters of vapor, vapor/liquid and liquid phase regions and 15 probability maps for each of the specific phase equilibrium method, were obtained. The analysis of the results confirmed the hypothesis that the use of Kohonen neural networks for separation of the classes was correct.

This work is licensed under a Creative Commons Attribution 4.0 International License

Keywords
physical properties, phase equilibrium, neural networks, neural network classification, learning procedure