In this study, differences in the chemical compositions of rebated excise duty diesel oil samples that were caused by fuel laundering were investigated. Two possible laundering pathways were simulated using either reduction or adsorption agents in model samples that were spiked with Solvent Yellow 124 and Solvent Red 19. The samples were characterized by their chromatographic fingerprints, which were recorded using gas chromatography coupled with a nitrogen chemiluminescence detector. The collections of fingerprints were further analyzed by discriminant partial least squares and the models with the optimal complexities presented the correct discrimination rates in the range of 69.1%-99.6%, respectively. The most informative fingerprint sections that were associated with the investigated differences were identified using the variable importance in projection, selectivity ratio and uninformative variable elimination methods. The reduced multivariate discriminant models presented a relatively high performance with the correct classification rates in the range of 74.9%-99.8%, respectively. O-toluidine and 2,5-diaminotoluene were identified as potential markers of diesel oil counterfeiting by laundering through a reduction agent.
原文链接
分享
引用
