一种检测溢油的快速分类辅助鉴别方法

A Fast Classification Auxiliary Identification Method for Detecting Spilled Oil

用气相色谱法和气相色谱-质谱法分析了原油样品的正构烷烃和甾萜烷生物标志物。针对在仪器分析过程中由各种因素带来的采集点漂移的现象,利用相关系数校正方法对保留时间进行了校正。实验结果表明,保留时间校正后大大提高了采集点的重合性。采用欧氏距离方法对校正后的色谱数据进行聚类分析,基于色谱数据的聚类分析对于降解油和非降解油的分类效果良好;而基于校正后甾、萜烷生物标志物数据的聚类分析结果能很好的将溢油样品与溢油源样品聚为一类,而与其他不相干油田的样品区分开来,聚类分析结果更能真实反映样品的真实信息。相关系数校正与快速聚类分析方法相结合为溢油来源的判断提供了一种快速分类辅助鉴别方法。

N-alkanes, steranes and terpanes of crude oil were analyzed by gas chromatography ana gas chromatography mass spectrometry ior oil spill identification, however, there is always drift of data col- lection point on chromatogram profiles, which makes the comparison between samples more difficult. To eliminate the drift of data collection point caused by variety of factors, correlation coefficient method was applied to warp the retention time to improve coincidence. It was proved that the superposition of data collection point was increased greatly after the retention time was warped using correlation coeffi- cient method. Warped chromatogram data was clustered with Euclidean distance. It was shown from the cluster result that cluster result based on gas-chromatogram data can distinguish the degraded oils and no degraded oils, cluster result based on terpanes and steranes data can cluster the spill oil samples and source oil samples as one class, and the other irrelevant samples were separated from the spill samples and source samples, the cluster results can reflect the real information of the samples. The combination of correlation coefficient warping and cluster analysis provided a fast classification auxiliary distinguish method for source identification of spill oil.