Abstract:

In order to accurately simulate the complex relationship between the molecular structure, molecular weight and boiling point of the compounds, 36 terpenoids in plant essence were studied in this essay. The topological index method was used to quantify the molecular structure, and artificial neural networks were applied to construct nonlinear models. The results show that the fitting accuracy of MSBPT with a 2:15:1 structure is 96%, and the prediction accuracy is 91%. The relationship between molecular structure and boiling point is strengthened by introducing molecular weight as an input variable. The artificial neural network is suitable for modeling the complex nonlinear relationship between the molecular structure of terpenoids and their boiling points, with high fitting and prediction accuracy. At the same time, the group contribution method has a wide application range and the topological index method has reliable data, which should be further studied in the field of forestry and agriculture.

Key words: artificial neural network, terpenoids, molecular structure, topological index, boiling point