Hybrid intelligence methods for modeling the diffusivity of light hydrocarbons in bitumen

Abstract

The solvent diffusivity is considered as a key factor in the design of solvent assisted processes in the bitumen field. In this study, a novel Adaptive neuro-fuzzy interference system (ANFIS) is employed to evaluate the diffusivity of the light hydrocarbons in the bitumen system. The particle swarm optimization (PSO) and genetic algorithm (GA) are adopted to promote ANFIS efficiency. The proposed models are established by a prepared dataset from multiple papers in the literature. Temperature (T), pressure (P) and molecular weight of alkanes (Mw) were considered as the input variables and on the other hand, Statistical parameters and graphical methods were used to appraise ANFIS, ANFIS-PSO, and ANFIS-GA performance. The results demonstrated that the highest correlation coefficient is related to ANFIS-PSO with R2 = 0.991 and 0.987 for train and test data, respectively. In the end, the results indicated that the ANFIS-PSO model has a higher level of desirability based on statistical parameters.

Keywords

• Chemical engineering
• Energy
• Petroleum engineering
• Multidisciplinary design optimization
• Artificial neural networks
• Diffusivity