Genetic algorithm for fuel spill identification

• Published in: Analytica chimica acta Vol. 437; no. 2; pp. 233 - 246
• Main Authors: Lavine, B.K, Brzozowski, D, Moores, A.J, Davidson, C.E, Mayfield, H.T
• Format: Journal Article
• Language: English
• Published: Amsterdam Elsevier B.V 27.06.2001; Elsevier
• Subjects: Analytical chemistry; Applied sciences; Chemistry; Chromatographic methods and physical methods associated with chromatography; Crude oil, natural gas and petroleum products; Energy; Exact sciences and technology; Feature selection; Fingerprint data; Fuel spill identification; Fuels; Gas chromatographic methods; Genetic algorithm; Machine learning; Pattern recognition analysis; Pollution caused by production, transportation and treatment of oil and oil shales. Water and soil pollution. Treatments. Pollution control; Pattern recognition analysis; Feature selection; Fuel spill identification; Genetic algorithm; Fingerprint data; Machine learning; Oil spill; Data analysis; Molecular structure; Evolutionary algorithm; Flame ionization detector; Pattern recognition; Multivariate analysis; Gas chromatography; Liquid fuel; Water pollution control; Environmental monitoring; Learning algorithm; Chemometrics; Principal component analysis
• ISSN: 0003-2670; 1873-4324
• DOI: 10.1016/S0003-2670(01)00946-1

Gas chromatography is frequently used to fingerprint fuel spills, with the gas chromatograms of the spill sample and the different candidate fuels compared visually in order to seek a best match. However, visual analysis of gas chromatograms is subjective and is not always persuasive in a court of law. Pattern recognition methods offer a better approach to the problem of matching gas chromatograms of weathered fuels. Pattern recognition methods involve less subjectivity in the interpretation of the data and are capable of identifying fingerprint patterns within gas chromatographic (GC) data characteristic of fuel-type, even if the fuel samples comprising the training set have been subjected to a variety of conditions. In this paper, we report on the development of a genetic algorithm (GA) for pattern recognition analysis of GC fuel spill data. The pattern recognition GA incorporates aspects of artificial intelligence and evolutionary computations to yield a “smart” one-pass procedure for feature selection. Its efficacy is demonstrated by way of two studies recently completed in our laboratory on fuel spill identification.