Abstract: Based on the composition characteristics of group II/III base oil, two kinds of neural network models was built with multiple neural network and radial basis probabilistic neural network. A nine parameter neural network model with the oxidation stability as the output variable was established taking the contents of straight chain alkanes, naphthenes, alkyl benzenes and viscosity index of the hydrotreated base oil as input variables. For the first time, the viscosity index was used as the input parameter of the prediction model, which greatly improves the prediction accuracy of the model. Based on the analysis of the factors affecting the oxidation stability of hydrotreated base oil, the alkane components of II/III base oil with positive/negative correlation with the oxidation stability were found. With the increase of linear alkanes and the decrease of dicyclic and tricyclic alkanes in II/III base oils with low aromatic, the oxidation stability of base oil increases.

Key words: lubricating oil, base oil, oxidation stability, neural network, artificial intelligence simulation