Influence of reading noise in Raman spectra: effect on the classification of adulterated type C gasoline using the Annotated Paraconsistent Logic of two values - LPA2v and Principal Component Analysis - PCA techniques Denis Medeiros dos Santos, Claudio Luiz Firmino, Dorotéa Vilanova Garcia, Landulfo Silveira Jr., Marcos Tadeu T. Pacheco

Abstract

Principal component analysis (PCA) and annotated paraconsistent logic of two values (LPA2v) were used to evaluate the influence of thermal noise and reading noise on the classification of adulteration in gasoline type C through the Raman spectrum. 15 samples of gasoline C and the respective Raman spectra were obtained. The Raman peaks of adulterants and naphtha were recognized in order to establish the pattern of gasoline with and without the presence of adulterants. The techniques showed an error of 0% (all samples were identified correctly) for the collection time of 10 s for the data analyzed via PCA and 5 s for the data analyzed via LPA2v. The techniques were shown to be efficient for the correct classification using spectra obtaining times greater than 5s.