Título: Clustering Crude Oil Samples Using Swarm Intelligence
Autores: Ferreira, Fernando;José de Seixas;Xavier, Gilberto;Ciodaro, Thiago;Torres, Alexandre
The identification of oil patterns in the distillation process provides useful information for the refinery operation and logistics. The a priori information concerning the characteristics expected by a given oil sample improves the logistic concerning which refineries should process the sample, together with pricing and marketing. This article presents the results of data mining models applied to a generic database of crude oil samples. Only information available in the beginning of the oil refinement process is used. Clustering techniques based on bio-inspired algorithms are applied to the data samples in order to extract structured patterns from data. Three algorithms were used: PSO, FSS and ABC. The silhouette index was used as the fitness function to be optimized. The results were later evaluated using other clustering quality index. The algorithms were able to find patterns beyond the standard oil classification, which considers only the oil density measure.
Oil and Gas;Swarm Clustering;Data Mining;Pattern Recognition
Código DOI: 10.21528/CBIC2017-64
Artigo em pdf: cbic-paper-64.pdf
Arquivo BibTeX: cbic-paper-64.bib