8290614318

7. Opublikowane badania własne

J Ontl et al/Futl 117(2014) 224-229

3. Results and dlscussion

The data chat was studied were first preprocessed according to the steps described in Section 2.3. In order to efficiently handle the Rayleigh scattering. the location of the peaks related to the scattering were detected (in all of the emission spectra registered). Then, fifteen sampling points corresponding to the Rayleigh scattering peaks were removed and interpolated using the procedurę described in Ref. [24J. The conected data were unfolded. aver-aged (as described in Section 2.3) and centered column-wise prior to further anatysis. The first two principal components described over 99.05% of the total data variance. The PCA score plot, which dispLays 60 diesel oil spectra on the PCI and PC2 piane, is

presented in Fig. 3. In order to investigate the scatter of the sam-ples induded in the tax groups being considered, they were marfced according to four possible discrimination problems (see Section 2.3).

Exploring the PC1 vs. PC2 score plot, two dusters of samples are observed. However, they do not represent any of the discrimination problems being considered. The score plot presented in Fig. 3a helps to spot separate dusters that match discrimination problem ni. Thus. the concentration of SY124 is a possible marker that can be used to indicate the process of illegal sorption. Further interpretation of the score plots with the other three discrimination criteria indicated (Fig. 3b-d) could not be associated with any particular duster of samples.

Coartalratioa |mj| * I. *|

SM 24    000022244555777?* 10 I* 10

•SB 19	82SI0 0J73	o 5 10 • i	7	3	« 0 5 10
		] •«•••«
	1
	I ____	1
1
		1		—

f

Fłjfc. 3. PCI vs PO scorr ploo of oil umptet markrd according to four discrimination schtmts (j)nl.(ta) n2. (c)n3 and (d) n4. Samptrs includrd in the rrluird ra* group Jrc maiWcd with 'U' and samples induded in the fuli ratę of dury group arr marbrd with 'O*.

Strona 79