Precipitation and deposition of asphaltene cause number of severe problems in downstream and upstream of oil industry. Hence, it is essential to present a dependable model for quantitative estimation of asphaltene precipitation. This paper aims to introduce a hybrid support vector regression (SVR) with harmony search (HS) as an intelligence approach to create quantitative formulation between amount of asphaltene precipitation and titration data. Harmony search is combined with SVR for determining the optimal value of its user-defined parameters. The optimization implementation by HS significantly improves the generalization capability of SVR. A dataset that includes 176 data points was employed in the current study, while 141 data points were utilized for constructing the model and the remainder data points (35 data points) were used for assessment of degree of accuracy and robustness. Evaluating the performance of constructed model based on statistical criteria indicates that the hybrid model has acceptable accuracy to estimate the amount of asphaltene precipitation from titration data. This study concludes that optimization of SVR with HS produces a smart hybrid model, which is an appropriate solution for modeling of the asphaltene precipitation.

No abstract