Study On Quantitative Identification Of Crude Oil By Multi-dimensional Chemical Fingerprinting

With the rapid development of science and technology and the full utilization of the earth's resources,large-scale oil exploitation activities have gradually emerged in the seas at home and abroad,and the probability of occurrence of offshore oil spills has also risen sharply.Therefore,it is important to analyze the leakage of oil and find efficient and rapid oil fingerprint identification methods for analyzing the types and sources of oil spills.In recent years,the identification technology of oil spills has received extensive attention from scholars at home and abroad.As far as the current situation is concerned,most of the identification techniques are still qualitative analysis by spectral comparison,but the correctness of qualitative analysis is highly subjective.The impact is blurred and the boundaries are blurred.The main problems in the technical methods of quantitative analysis are:lack of variable types,inadequate model establishment,etc.In order to solve the related problems,this study starts from the representative characteristics of crude oil products,and builds a comprehensive multi-dimensional chemical fingerprint comprehensive model from point to point,aiming at efficiently identifying Middle East and non-Middle East crude oil.In this paper,16 kinds of oil samples from 8 Middle Eastern crude oils and 8 non-Middle East crude oils were studied.The oil characteristics of all oil samples before and after weathering were investigated.The principal component analysis of n-C16-n-C35 of normal paraffins was carried out.The cumulative interpretation contribution rate of n-C16-n-C20 is 91.8%.The oil samples were scanned at ?X=30 nm by constant-wavelength synchronous fluorescence method.The crude oils showed fluorescence peaks at 280±2 nm,302±2 nm,332±2 nm,and 380±2 nm.Using the n-C16?n-C20 five principal component eigenvalues of n-alkanes,four fluorescence eigenvalues at 280±2nm,302±2nm,332±2nm,and 380±2nm wavelengths and Ni/V to build three Fisher discriminant models.The correct rates of the three model training sets were 81.25%,81.25%,and 68.75%,respectively.After the weathering for 30 days,the oil samples were used as verification.The accuracy of the verification set was 87.5%,75%,and 75%,respectively.Analyze the correct rate,summarize the limitations,establish a multi-dimensional chemical fingerprint identification model,adopt the variable circulation screening method,exhaust all the variables,cross-contrast data,select the best,and finally determine n-C16?n-C18?n-C27?n-C29,n-C22 ? n-C31,332±2nm,380 ±2nm eigenvalue and the Ni/V value are used as variables.At this time,the correct combination rate of variables is higher,the correct rate of the final model training set is 100%,the correct rate of the verification set can reach 93.75%,and the accuracy of identifying crude oil is effective improve.