Ability of analytical and artificial approaches for prediction of the volumetric properties of some polymer blends

In this research the statistical–mechanical equation of state and artificial neural network approaches are developed for volumetric properties of polymer blends. The Ihm–Song–Mason (ISM) and Tao–Mason (TM) equations of state based on density and surface tension at melting point (ρm and γm), as scaling constants, were developed. The calculation of second virial coefficients (B2), effective van der Waals co-volume (b) and correction factor (α) are required for judgment about applicability of these models. Also another model such as an artificial neural network (ANN) based on back propagation training with 15 neurons was used. A collection of 5397 data points for five polymer blends in the temperature range of 298.15–605.05 K and pressures to 200 MPa was used. The obtained results by ISM, TM and ANN models had good agreement with the experimental data with absolute average deviations of 0.72%, 0.84% and 0.34%, respectively. Among the applied methods, higher efficiency and better accuracy was achieved using ANN.