Radial basis function (RBF) network for modeling gasoline properties

摘要

One of the important products of a crude oil refinery is gasoline which is greatly used as a liquid fuel. Hence, it is necessary to accurately specify its quality by measuring different properties of gasoline. In this study, radial basis function neural networks were utilized for estimation of different characteristics of gasoline including specific gravity (SG), Reid vapor pressure (RVP), research octane number (RON) and motor octane number (MON). The genetic algorithm was used as an optimization algorithm to optimize the maximum neuron number and spread of model. Results reveal that the developed GA-RBF model is effective and precise for estimating experimental data. Furthermore, comparison between the GA-RBF model and a previously reported LSSVM model in literature shows the superiority of GA-RBF model.