The Logistic Regression Based On Diagnostic Ratio For Identification Of The Middle East Crude Oil

At present,most of the oil spill identification is the qualitative spectrum research,but the research on the quantification and identification of oil spills is less.A Logistic regression model was established to quantify the identification of Middle East crude oils and non-Middle East crude oils in this study and make the tracing results more intuitive and clear.At the same time,it provides a technical support to accurately determine the type of weathering oil spills on the sea.In this paper,seven types of Middle East crude oils and thirteen types of non-Middle East crude oils were studied and analyzed by GC-FID.And the influence of weathering on distribution of n-alkanes and diagnostic ratio of crude oils was discussed.Then the similarity analysis and cluster analysis were used to select the modeling parameters to establish Logistic regression model for distinguishing the Middle East crude oils.The n-alkanes of Middle East crude oils are mainly concentrated on n-C8?n-C17 and n-C3 is the boundary points.The content of n-C8?n-C13 present a rising tendency and decrease after n-C13,but the distribution trend was similar overall.The distribution of n-alkanes of non-Middle East crude oils is quite different.Most n-alkanes of non-Middle East crude oils are focus on n-C13?n-C25,small of them contain abundant light components.There is major differences between Middle East crude oils and non-Middle East crude oils in aspects of diagnostic ratio.After 30 days of weathering,all the n-alkanes before n-C13 are completely lost.But the diagnostic ratios of n-C17/Pr,n-C18/Ph and Pr/Ph are essentially unaffected by weathering so that it can still be used as the gist for identification of weathering oil spills.The similarity between crude oils was analyzed by using n-C7/Pr,n-C18/Ph and Pr/Ph as eigenvectors.Similarity between the Middle East crude oil is higher,the similarity between Middle East crude oil and Non-Middle East crude oil on the low side.What show that there is a significant difference between the Middle East crude oils and non-Middle East crude oils so it can be modeled and analyzed as a dichotomous events.In comparison with n-alkanes,the diagnostic ratio as characteristic variable,it can be used to classify the Middle-East crude oils and non-Middle East crude oils more accurately.So n-C17/Pr,n-C18/Ph and Pr/Ph are selected as modeling parameters.The Logistic regression model was established with the the parameters of diagnostic ratio and the discriminating results of the oil species was highly consistent with the predicted targets.The example validation results of the weathered oil samples and the literature datas show that the model of accuracy rates were 100%and 95%respectively for classification of the Middle East crude oils and non-Middle East crude oils.Indicating that the oil species Logistic model is not only suitable for the identification of unweathered crude oils,but also for short-term weathering crude oils.