DefinePK

Abstract

In this research, the ability of multi-layer perceptron neural networks to estimate vapour liquid equilibrium data have been studied. Four binary systems (R1270+R290, CO₂+R290, R125+R290, and R32+R290) have been investigated in the large ranges of temperature and pressure. These systems show different deviations from the Raoult's law. The networks with one hidden layer consist of five neurons are developed as the optimal structure. The networks were trained and then used as one-step tools without any iteration to estimate VLE data. For these binary systems, uncertainties in the ANNs results were not more than 0.126, 0.371, 0.221, and 0.613 %, respectively. In addition, the abilities of ANNs are shown by comparisons with Margules, van Laar, and some other usual correlations. Results show capability of presented networks obviously.