|
Abstract
|
Two different modeling methods have been proposed to relate the Peng-Robinson-Stryjek-Vera (PRSV) parameter, κ1, to some common thermodynamic constants, including critical temperature (Tc), critical pressure (Pc), acentric factor (ω) and molecular weight (Mw). The methods are artificial neural network (ANN) and adaptive networkbased fuzzy inference System (ANFIS). A set of 159 data points (116, 23 and 20) was used for construct training, validating and testing, respectively. The radius parameter of ANFIS was determined through genetic algorithm (GA) optimization technique. The ANN and especially ANFIS results are in a good agreement with most of the compound groups.
|