Fuel Spill Identification by Gas Chromatography - Genetic Algorithms/Pattern Recognition Techniques

Abstract

Gas chromatography and pattern recognition methods were used to develop a potential method for typing jet fuels so a spill sample in the environment can be traced to its source. The test data consisted of 256 gas chromatograms of neat jet fuels. 31 fuels that have undergone weathering in a subsurface environment were correctly identified by type using discriminants developed from the gas chromatograms of the neat jet fuels. Coalescing poorly resolved peaks, which occurred during preprocessing, diminished the resolution and hence information content of the GC profiles. Nevertheless, a genetic algorithm was able to extract enough information from these profiles to correctly classify the chromatograms of weathered fuels. This suggests that cheaper and simpler GC instruments can be used to type jet fuels.