Molecular modeling of the ideal gas enthalpy of formation of hydrocarbons

The ideal gas enthalpy of formation (HFOR) plays a key role in energy balance calculations and the investigation of bond energies, resonance energies and the nature of chemical bonds. Development of accurate structure-based estimation methods for HFOR of a large variety of chemical species can enhance the capability of process and product. In this work, quantitative structure–property relationship (QSPR) study was performed to simply investigate the HFOR for a diverse DIPPR dataset include 1783 hydrocarbons from 80 diverse chemical classes. Based on the multivariate linear regression (MLR) a multivariate model was developed using robust binary particle swarm optimization (PSO) for the feature selection step. Next, based on the scaled variable reduced coordinates (SVRc), a novel robust mathematical modeling strategy was introduced using PSO that was successfully implemented and checked. The results of the PSO-MLR model and PSO-SVR illustrated that R2 between predicted and experimental values were 0.9500 and 0.9644, respectively.