Artificial Neural Network and Attenuated Total Reflectance-Fourier Transform Infrared Spectroscopy to identify the chemical variables related to ripeness and variety classification of grapes for Protected. Designation of Origin wine production

[Display omitted] •FTIR analysis in grape skin allowed identifying sample variety using ANNs.•ATR fast analysis without pretreatment avoids undesired structural changes of sample.•Compounds affecting ripeness and classification were identified with Olden’s method.•Pectin, polysaccharides and specially fructose, have the strongest influence. The vineyard grown in the territories included in the Protected Designations of Origin (PDO) classification of the European Union, present unique organoleptic properties of colour, aroma and flavour. Development of techniques for identifying grape varieties or ripeness among other characteristics, are key interesting for the PDO control and quality. Attenuated total reflectance (ATR) allows fast recording spectra without sample pre-treatment, thus avoiding undesired physical and/or chemical changes of the sample. This method works in a rapid, non-destructive and easy-to-use way. The fast-fourier transform infrared spectroscopy (FTIR) analysis of five grape varieties (Alabarín blanco, Mencía, Verdejo negro, Albarín negro and Carrasquín) used for wine production of PDO Vino de Cangas provided information enough for the identification of grape class using artificial neural networks (ANN). Despite the statistical similitude of the FTIR spectra among different grapes and maturity state, ANN resulted to be a helpful tool for classifying grape samples according to the variety or to their ripeness degree. Furthermore, compounds present in grapes that can most influence such classification can be outlined from the ANN. In this context, pectin and polysaccharides are especially significant in variety and ripeness identification, whereas polyphenols and fructose provide useful information for ripeness degree classification of grapes.