The Study On Digital Identification Of Sunken Oil During Long-term Weathering

In the process of oil exploration,development,refining and transportation,oil may leak and flow to water surface or beach,etc.,thus forming a film of uneven thickness.This phenomenon is known as oil spill.At present,the oil spill treated standard is invisible oil film on the water surface,but in fact oil may be sunken and submerged due to various reasons.Since sunken and submerged oil is not easily to be found,usually produces unpredictable effect,and its floating will bring unprepared environmental hazards and economic losses.Therefore,it is necessary to identify sunken and submerged oil,in order to improve the supervision of oil spill accident at sea.In this paper,sunken oil in the sunken and submerged oil is set as the identified object.Firstly,1 kind of crude oil(A)and 2 kinds of fuel oils(B and C)of 340 days weathering are analyzed by gas chromatography(GC-FID).Secondly,repeatability limit method is used to screen out sable diagnostic ratios of floating oil on water and sunken oil under water.Finally,sunken oil is distinguished by 3 digital identification methods based on these diagnostic ratios,such as multidimensional scaling analysis,cluster analysis and discriminant analysis.The main results of this paper are as following:(1)The screening results showed that the stability of diagnostic ratios is not only related to oil types and weathering time,but also closely related to existence forms of spill oil.It appears as that the effect of weathering on sunken oils of A and B with relatively more light components are greater than their floating oils,but C with more heavy components is just the opposite.The research eliminate all unstable diagnostic ratios,and finally get 4 identification indexes suitable for floating oil and sunken oil of long-term weathering,respectively are(C19+C20)/(C19+C20+C21+C22),C17/(C17+Pr),CPI1,and OEP1.(2)The digital identification results indicate that multidimensional scaling analysis using 5 kings of metrics cannot fully identify the sunken oil of long-term weathering.However,cluster analysis using ward method and squared Euclidean distance and Bayes discriminant analysis not only are able to distinguish different oil samples,but also cluster 3 kinds of sunken oils and their sources successfully.It is reasonable to use clustering analysis based on stable diagnosis ratio and Bayes discriminant analysis to identify the sunken oil by the preliminary judgment of oils in this experiment.They can reliably trace the sunken oil in outdoor simulated weathering after 340 days.The above 2 kinds of methods prove each other,which further enhance the reliability of the identification results of sunken oil.