On the prediction of chemical exergy of organic substances using least square support vector machine

ABSTRACT

Knowledge of a technique which could significantly predict standard chemical exergy values seems to be beneficial in chemical processes. In this regard, a combination of genetic algorithm (GA) and least square support vector machine (LSSVM) was developed. A set of 134 data was utilized for model development from which 114 and 20 data were allotted to training and testing phases, respectively. In addition, the average absolute relative error and coefficient of determination values were 1.8351 and 0.99848, respectively. In this study, molecular weight of substance, atomic polarization, and number of atoms in any substances were selected as input data to introduce the subjected substances and consequently, the chemical exergy was selected for target of prediction. In conclusion, it is comprehensively established that the proposed model is considerably efficient due to its highly accurate prediction and GA-LSSVM introduced as an efficient model for the prediction of chemical exergy of organic substances.

KEYWORDS:

• Chemical exergy
• genetic algorithm
• LSSVM
• molecular structure
• organic substances