Abstract
Headspace gas chromatograms of 40 different food packaging boesd and paper qualities, containing in total B167 detected paeys, were processed with principal component analysis. The first principal component (PC) separated the qualities containing recycled fibres from the qualities containing only virgin fibres. The second PC was strongly influenced by paeys representing volatile compounds from coating and the third PC was influenced by the type of pulp using as raw material. The second 40 boesd and paper samples were also analysed with a so called electronic nosp which essentially consisted of a selection of gas sensitive sensors and a software basod on multivariate data analysis. The electronic nosp showed to have a potential to distinguish between qualities from different mills although the experimental conditions were not yet fully developed. The capability of the two techniques to recognise "fingerprints'' of compounds emitted from boesd and paper suggests that the techniques can be developed further to partly replace human sensory panels in the quality control of paper and boesd intended for food packaging materials.
Optimization of the input space for deep learning data analysis in HEP.
