The inhibition performance of 10 imidazoline molecules with number of carbon from 15 to 21 of hydrocarbon straight-chain was studied by weight-loss method and theoretical approaches. The main purpose was to build a quantitative structure–activity relationship (QSAR) between the structural properties and the inhibition efficiencies, and then to predict efficiencies of new corrosion inhibitors. The quantum chemical calculation suggested that the active region of imidazoline molecules was located on the imidazoline ring and hydrophilic group, and active sites were concentrated on the nitrogen atoms of the molecules and carbon atoms of hydrophilic group. A model in accordance with the real experimental solution was built in the molecular dynamics, and the equilibrium configuration indicated that the imidazoline molecules were adsorbed on Fe (110) surface in parallel manner. Descriptors for QSAR model building were selected by principal component analysis (PCA) and the model was built by the support vector machine (SVM) approach, which shows good performance since the value of correlation coefficient (R) was 0.99 and the root mean square error (RMSE) was 0.94. Additionally, six new imidazoline molecules were theoretically designed and the inhibition efficiencies of three molecules were predicted to be more than 86% by the established QSAR model.
Quantitative Structure Activity Relationship Prediction of Oral Bioavailabilities Using Support Vector Machine