Evolving new group contribution-LSSVM model to estimate standard molar chemical exergy of pure organic substances

Abstract

Chemical exergy values of pure organic compounds are required in order to perform an exergy analysis to achieve the optimum conditions. Development of reliable predictive tools for standard molar chemical exergy estimation, is of great importance. A least squares support vector machine (LSSVM) based group contribution (GC) method is proposed for standard molar chemical exergy prediction of pure organic compounds. The proposed model is trained and evaluated based on a comprehensive data base comprising standard molar chemical exergy for 133 organic compounds. 47 chemical substructures are employed in the process of model development. The proposed model is evaluated using different graphical and statistical error analysis. Determination coefficient (R2) and average absolute relative deviation (AARD%) values of 1.00 and 0.56% indicate the applicability potential and reliability of the predictions from the proposed model.