Instrumental measurement of wine sensory descriptors using a voltammetric electronic tongue
The approach presented herein reports the application of a voltammetric electronic tongue (ET), in contrast with a wine tasting sensory panel, as a tool for standardized wine tasting; concretely, to achieve the discrimination of different wine DOs (Denominación de Origen, a mark related to its geogr...
| Autores: | , , , , |
|---|---|
| Tipo de recurso: | artículo |
| Fecha de publicación: | 2015 |
| País: | España |
| Institución: | Universitat Autònoma de Barcelona |
| Repositorio: | Dipòsit Digital de Documents de la UAB |
| Idioma: | inglés |
| OAI Identifier: | oai:ddd.uab.cat:144738 |
| Acceso en línea: | https://ddd.uab.cat/record/144738 https://dx.doi.org/urn:doi:10.1016/j.snb.2014.09.081 |
| Access Level: | acceso abierto |
| Palabra clave: | Electronic tongue Linear discriminant analysis Artificial neural network Voltammetric sensor Wine Sensory panel |
| Sumario: | The approach presented herein reports the application of a voltammetric electronic tongue (ET), in contrast with a wine tasting sensory panel, as a tool for standardized wine tasting; concretely, to achieve the discrimination of different wine DOs (Denominación de Origen, a mark related to its geographical region and ensuring high-quality levels) and the prediction of the global score assigned by the trained sensory panel. To this aim, a voltammetric array of sensors based on metallic and bulk-modified graphite electrodes was used as the sensing part, while chemometric tools such as linear discriminant analysis (LDA) and artificial neural networks (ANNs) were used as the qualitative and quantitative modelling tools. Departure information was the set of voltammograms, which were first preprocessed employing fast Fourier transform(FFT), followed by removal of less-significant coefficients employing a stepwise inclusion method and pruning of the inputs. The trend, in global scores, was modelled successfully with a 92.9% of correct identification for the qualitative application, and a correlation coefficient of 0.830 for the quantitative one (with 14 and 20 samples for the external test subsets, respectively). |
|---|