Aromatic stability and degradation of Extra Virgin Olive Oil

The Panel Test (PT) is performed by certified assayers and recognized by European Union. From a statistic point of view, it is a methodology sufficiently objective and repetitive so realizing a Metrology. The aim of this work is the study of extra virgin oil olive degradation and stability. The method uses a cascade of two neural networks which inputs are an essential set of quality parameters coming from the Panel Tests made on extra virgin olive oils samples and gas chromatographic analysis on them. Nine extra virgin olive oil samples have been examined. The results of the PT are classified by an unsupervised neural network and related on a Kohonen Map. Then, they are correlated to Gas Chromatographic analysis of olive oil by the Multi Layer Perceptron with Back-Propagation (MLP-BP), a supervised neural network. The MLP-BP is able to classify a sample different from those used for the learning. The olive oil degradation is analyzed by comparing the results on the same sample performed in different periods. Instead, the aromatic stability is observed by comparing different sample from the same producer. The degradation has been observed by the change of the intensity of the signals: the reduction of aromas "fruity" and "green grass" is detected, typical flavors of freshness of the product, in in some oils the flavors "spicy" and "mint" appear. As regards the aromatic stability, many products have maintained the bearing characteristics while the few variations are related to the disappearance of aromas of flowers and/or apple.