Abstract:A method is presented for the correlation of the physical properties of oils using BP neural network. The chemical composition of heavy oils are then correlated with the ordinary physical properties such as density, viscosity, C/H ratio and Conradson carbon residue etc.. The results are better than those obtained from regressive analysis method in literature. The BP neural network method is characterized by its universal mathematical model, automatic formation of correlation relation between physical properties and correlation parameters, simplicity and convenience in calculation, and is appropriate for the oil systems with very complicated components.

Correlation of chemical composition of heavy oils using BP neural network