Electronic tongues to assess wine sensory descriptors

Xavier Cetó; Andreu González-Calabuig; Nora Crespo; Sandra Pérez; Josefina Capdevila; Anna Puig-Pujol; M Del Valle

doi:10.1016/j.talanta.2016.09.055

Electronic tongues to assess wine sensory descriptors

Talanta. 2017 Jan 1:162:218-224. doi: 10.1016/j.talanta.2016.09.055. Epub 2016 Sep 24.

Authors

Xavier Cetó¹, Andreu González-Calabuig², Nora Crespo², Sandra Pérez³, Josefina Capdevila⁴, Anna Puig-Pujol⁴, M Del Valle⁵

Affiliations

¹ Future Industries Institute, University of South Australia, 5095 Adelaide, Australia.
² Sensors and Biosensors Group, Department of Chemistry, Universitat Autònoma de Barcelona, Edifici Cn, 08193 Bellaterra, Barcelona, Spain.
³ Future Industries Institute, University of South Australia, 5095 Adelaide, Australia; Sensors and Biosensors Group, Department of Chemistry, Universitat Autònoma de Barcelona, Edifici Cn, 08193 Bellaterra, Barcelona, Spain.
⁴ Estació de Viticultura i Enologia, INCAVI-IRTA, 08720 Vilafranca del Penedès, Spain.
⁵ Sensors and Biosensors Group, Department of Chemistry, Universitat Autònoma de Barcelona, Edifici Cn, 08193 Bellaterra, Barcelona, Spain. Electronic address: manel.delvalle@uab.cat.

PMID: 27837821
DOI: 10.1016/j.talanta.2016.09.055

Abstract

This work reports the application of an electronic tongue as a tool towards the analysis of wine in tasks such as its discrimination based on the maturing in barrels or the prediction of the global scores assigned by a sensory panel. To this aim, red wine samples were first analysed with the voltammetric sensor array, without performing any sample pretreatment. Afterwards, obtained responses were preprocessed employing fast Fourier transform (FFT) for the compression and reduction of signal complexity, and obtained coefficients were then used as inputs to build the qualitative and quantitative models employing either linear discriminant analysis (LDA) or partial least squares regression (PLS), respectively. Satisfactory results were obtained overall, with a classification rate of 100% in the discrimination of the type of barrel used during wine maturing, a normalized NRMSE of 0.077 in the estimation of ageing time (months) or 0.11 in the prediction of the scores (0-10) from a trained sensory panel (all for the external test subset).

Keywords: Ageing; Electronic tongue; Partial least squares regression; Sensory panel; Voltammetric sensors; Wine.