| Literature DB >> 27410901 |
Ruifang Yang, Nanjing Zhao, Xue Xiao, Gaofang Yin, Shaohui Yu, Jianguo Liu, Wenqing Liu.
Abstract
It is still difficult to determine the concentrations of polycyclic aromatic hydrocarbons accurately in natural water by fluorescence technique because of their low solubility, different fluorescent intensity, and the complex interferents from water environments. In this work, three-way fluorescence spectra combined with three methods including three-way parallel factor analysis, multi-way partial least square with residual bilinearization and unfolded partial least square with residual bilinearization were used to predict the concentrations of polycyclic aromatic hydrocarbons at the μg L<sup>-1</sup> level in reservoir and river water, respectively. The prediction abilities of these methods on different analytes were evaluated by validation sets. The results demonstrate that unfolded partial least square with residual bilinearization yields the optimal results with relative error less than or equal to 6% for phenanthrene, pyrene, anthracene and fluorene, and 35% for acenaphthene and fluoranthene in different water backgrounds.Entities:
Year: 2016 PMID: 27410901 DOI: 10.1364/OE.24.0A1148
Source DB: PubMed Journal: Opt Express ISSN: 1094-4087 Impact factor: 3.894