Critical Properties and Acentric Factors of Pure Compounds Modelling Based on QSPR-SVM with Dragonfly Algorithm

Abstract

This work aimed to model the critical pressure, temperature, volume properties, and acentric factors of 6700 pure compounds based on five relevant descriptors and two thermodynamic properties. To that end, four methods were used, namely, multi-linear regression (MLR), artificial neural networks (ANNs), support vector machines (SVMs) using sequential minimal optimisation (SMO), and hybrid SVM with Dragonfly optimisation algorithm (SVM-DA) to model each property. The results suggested that hybrid SVM-DA had better prediction performance compared to the other models in terms of average absolute relative deviation (AARD%) of {0.7551, 1.962, 1.929, and 2.173} and R 2 of {0.9699, 0.9673, 0.9856, and 0.9766} for critical temperature, critical pressure, critical volume, and acentric factor, respectively. The developed models can be used to estimate the property of newly designed compounds only from their molecular structure. Cilj ovog rada bio je modeliranje kritičnog tlaka, temperature, volumnih svojstava i acentričnih čimbenika 6700 čistih spojeva na temelju pet relevantnih deskriptora i dva termodinamička svojstva. U tu svrhu primijenjene su četiri metode: višestruka linearna regresija (MLR), umjetna neuronska mreža (ANN), metoda potpornih vektora (SVM) i algoritam optimizacije Dragonfly (SVM-DA), koji se za modeliranje svakog svojstva koriste sekvencijalnom minimalnom optimizacijom (SMO) i hibridnim SVM-om. Rezultati su pokazali da hibridni SVM-DA daje bolje predviđanje u odnosu na ostale modele u smislu postotka prosječnog apsolutnog relativnog odstupanja (AARD%) od {0,7551, 1,962, 1,929 i 2,173} i R 2 od {0,9699, 0,9673, 0,9856, i 0,9766} za kritičnu temperaturu, kritični tlak, kritični volumen i acentrični faktor. Razvijeni modeli mogu se primjenjivati za procjenu svojstava novodizajniranih spojeva samo iz njihove molekularne strukture.