Prediction of Vapor-liquid Equilibrium Databy Using Radial Basis Neural Networks

Most of the Chemical Engineering processes are nonlinear and complex in nature. They often require conventional modeling and simulation techniques based on certain simplified transport, kinetic and thermodynamic assumptions. These assumptions may, however, alter the exact nature of the system and would provide misleading information about the complex behavior of the system. An artificial neural network has the ability to overcome these limitations of the conventional approach by extracting the desired information directly from the data. In this paper radial basis network, a new generation of artificial neural network, has been successfully incorporated for the prediction of vapor liquid equilibrium data for binary systems including two azeotropes and a ternary system. Radial basis networks require lesser neurons than standard feed forward backpropagation and they can be trained in a fraction of time. From this work it is been proved that radial basis neural network has been successfully used for the prediction of vapor liquid equilibrium (VLE) data.