|
| Concentration Dependent Synchronous Fluorescence Oil Spill Fingerprinting Identification Based on Principal Component Analysis and Support Vector Machine |
| |
| View Full Text Download reader |
| DOI：10.3969/j.issn.1004-4957.年份.月份 |
| KeyWord:concentration dependent synchronous fluorescence spectroscopy concentration-synchronous-matrix-fluorescence spectroscopy(CSMF) oil spill fingerprinting principal component analysis(PCA) support vector machine(SVM) |
| Author | Institution |
| WANG Chun-yan,SHI Xiao-feng,LI Wen-dong,REN Wei-wei,ZHANG Jin-liang* |
1.北京师范大学资源学院;2.潍坊学院物理与电子科学学院;3.中国海洋大学光学光电子实验室 |
|
| Hits: 2604 |
| Download times: 1054 |
| Abstract: |
| In this paper,Concentration-Synchronous-Matrix-Fluorescence Spectroscopy(CSMF) was applied to characterize the chemical fingerprint information more comprehensively by adding concentration as a new dimension to fluorescence spectroscopy.Two tiered petroleum related sample sets(including the different spill oil types and different source crude oil sample set,and the closely-related source sample set with disturbance of weathering and seawater adulteration) were analyzed by principal component analysis(PCA).The results showed that for the crude oil samples from China,the weathering have no significant effect on the CSMF,and the PCA can classify the samples into different oil types in the principal components space according to the oil heaviness.Support Vector Machine(SVM),along with Leave-One-Out Cross-Validation,was used for confirmation of the validity of this method.100% accuracy was obtained for the different spill oil types and different source crude oil sample set,and 77% accuracy was for the closely related source sample set with disturbance of weathering and seawater adulteration.Detailed discussion indicated that pair-wise classification,can acquire higher accuracy than multi classification,and a tiered classification method from multi classification of different oil spill types to pair wise classification of closely related crude oil is then recommended for oil species identification.All the results suggested that the CSMF can be used as a rapid and reliable detection and characterization method for petroleum oil contaminants. |
| Close |
|
|
|