A digital identification of GC-FID oil fingerprints based on wavelet decomposition and fractal cluster analysis

A digital identification method of GC-FID oil fingerprint has been developed, which is based on the wavelet decomposition, fractal dimension calculation, and cluster analysis. The procedure of the identification consists of three steps: at first to separate the chromatographic curve into different series with different frequencies by the wavelet analysis, then to calculate the values of the fractal dimensions of the series curves respectively. After that, to do cluster analysis on the fractal data, according to the results of the cluster analysis, the oil samples can be compared and classified recognized. To apply the method on a few oil spill events, the results showed that the method can do better digital identification on shapes of GC-FID fingerprint curves of the oil spill samples. It can remarkably reduce number of the oil samples that need to be further identified. During the procedure, it is the key function that the fractal dimension can numerically describe well details of the curve shape. The supplementary method can be a good objective tool for oil spill identification.