| 摘要: |
| 于1986年对我国常见的14种油品和1984年8月大连湾海面溢油进行气相色谱分析，用模糊数学方法建立ECCIM贴近度聚类分析的海面溢油鉴别模式（此处公式见PDF文件），提出油种鉴别的Euclid贴近度置信水平λi的模糊区概念。油种及其海面溢油的Euclid贴近度聚类分析动态聚类图表明，油种鉴别的置信水平λi主要受风化和实验误差的影响，大庆原油、30＃重柴油、20＃重柴油、0＃轻柴油以及任丘原油各自分别同其它13种油可分类的置信水平λi的范围依次为：[0.8726，0.9953），[0.8726，0.9970），［0.8726，0.9857），[0.8807，0.9902）和[0.8726，0.9907）。 |
| 关键词: Euclid贴近度 模糊聚类分析 鉴别 溢油 |
| DOI： |
| 分类号: |
| 基金项目: |
附件 |
|
| CLUSTER ANALYSIS OF EUCLID APPROACH FOR IDENTIFICATION OF OIL SPILL AT SEA SURFACE |
|
Xu Hengzhen,Shang Longsheng,Zhou Chuanguang,Liu Tong
|
|
Institute of Marine Environmental Protection, SOA, Dalian 116023
|
| Abstract: |
| Fourteen common oil types of China and spilled oils at sea surface in Dalian Bay were analysed with gas chromatograms in 1984 - 1986. The distributions of hydrocarbons of nine crude oils and five fuel oils mostly ranged from nC8 - nC32 and nC10 - nC24. The impact of weathering on the oil chromatogram fingerprint is important. One month weathering experiment on 30# heavy diesel oil, 0# light diesel oil, and Renqiu crude oil on the seawater, showed the fluctuation of the weathered oil with carbon less than nC16 was large, than the fluctuation of the weathere oil with Carbon more than nC16, especially Pr and Ph, was less. Pr/nC17, Ph/nC18, Pr/Ph and (nC21+nC22) / (nC19+nC20) oil type information points for data handling. A model is established for fuzzy mathematics clustering analysis of Euclid approach for identification of sea surface spilled oil. The Euclid approach between the types of oil is expressed as:
(For the equations please see the PDF file.)
where xik is the ratio of the hydrocarbon content k in the oil i, xjk is the ratio of the hydrocarbon content k in the oil j. If the spilled oil and the suspected source oil come from the same source, their Eucllid approach will be equal to or near 1. The concept of fuzzy area of confidence level λi of Euclid approach for identifying sypes of oil was introduced. The confidence level λi for the identification of the tea surface spilled oil is found to be greatly influenced not only by the types of the oil, but also by weathering and experimental error. In geneal, the threshold value of this effects is equal to or more than 0.9857. The results show that the confidence levels λi for differentiating Daqing crude oil, 30# heavy diesel oil, 20 heavy diesel oil, 0# light diesel oil, and Renqiu crude oil, from other thirteen types of oil, are [0.8726, 0.9953), [0.8726, 0.9970), [0.8726, 0.9857),[0.8807, 0.9902) and [0.8726, 0.9907), respectively. |
| Key words: Euclid approach, Fuzzy clustering analysis, Identification, Spilled oil |
