Journal of Emerging Trends in Engineering and Applied Sciences: Research Areas

Abstract:
In an increasingly technological era, the dependence on oil and gas is on the increase. In order to meet this pressing need, oil exploration with associated spillage have been increasing at alarming rates, which in turn endangers the environment and generate huge oil spillage data repositories. Efforts to deal with oil spillages are inefficient because there is no clear knowledge and understanding of the relationships and inter-dependencies among oil spillage attributes. The purpose of this work is to discover patterns, rules and relationships among oil spillage attributes for oil spillage risks analysis. A hybrid methodology based on neural network, genetic algorithm and Fuzzy Logic for the discovery of patterns in terms of relationships, rules and inter-dependencies from oil spillage dataset is proposed. The system was implemented using 11Ants Analytics and Matlab using data on oil spillage incidences associated with oil exploration activities. A total of 33 comprehensible rules were discovered, the performance of the rules on the test data yielded an average accuracy and precision of 98.68% and 97.90% respectively. The rules are therefore adequate and suitable for the estimation of the oil spillage magnitude and can be deployed into the knowledge base of an expert system for oil spillage risk analysis. This knowledge will therefore guide human experts in the reduction of oil spillage incidences and associated risks.

Keywords: pattern discovery, data mining, oil spillage, risk analysis, rules.

Download full paper