Aplicação da técnica de espectroscopia do visível e infravermelho próximo (VIS/NIR) no controle físico-químico da qualidade de vinhos produzidos no submédio do Vale do São Francisco

• Ano de defesa: 2020
• Autor(a) principal: Barreto, Naiane Santos
• Orientador(a): Marques, Aline Telles Biasoto
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: por
• Programa de Pós-Graduação: Pós-Graduação em Ciência e Tecnologia de Alimentos
• Palavras-chave em Português: Tecnologia de alimentos; Vinho e vinificação; Vinhos espumantes; Uva; Cultivo de uva; Rio São Francisco; Vitis vinífera L.; Vinho tropical; Espumante; Regressão por mínimos quadrados parciais (PLS)
• Palavras-chave em Inglês: Tropical wine; Sparkling wine; Partial least squares regression (PLS)
• Área do conhecimento CNPq: CIENCIAS AGRARIAS::CIENCIA E TECNOLOGIA DE ALIMENTOS
• Link de acesso: https://ri.ufs.br/jspui/handle/riufs/13899

The quality control of wines and sparkling wines is carried out in wineries by means of specific physical-chemical analyzes, most of which are provided for in legislation to determine standards of identity and product quality. They are often based on time-consuming analytical techniques that require expensive equipment, sample preparation and / or use reagents. In this sense, as an alternative to traditional analysis methodologies, spectroscopy in the visible and near infrared region (Vis / NIR) is one of the analytical techniques that have been proposed for the evaluation of wines, mainly due to its simplicity, speed (analysis in 20 seconds), versatility, reliability and low cost. The aim of this study was to develop multivariate calibration models using a portable Vis / NIR spectrometer to determine quality parameters in three groups of fine wines of the Vitis vinífera L. varieties produced in the Sub-Middle region of the São Francisco Valley, they are: red wines containing 198 samples, white wines containing 86 samples, sparkling white wines containing 35 samples, and pink sparkling wines containing 59 samples; aiming at the total or partial replacement of conventional analytical methods. Additionally, the previous concentration of wine samples was tested using the DESIR technique, to obtain models with greater reliability for the prediction of substances present in low concentration in wines, being evaluated the contents of total phenolic compounds and monomeric anthocyanins. The multivariate calibration models were developed by PLS regression, using the SNV technique for spectral pre-processing. Models with satisfactory predictive performance were built for the determination in red wines ( Group I) of pH, alcohol content and h (hue angle or hue). For white and sparkling white wines (Group II ad III), the parameters density, alcohol content, dry extract, coordinate b *, L * (luminosity), C * (Chroma or saturation) and angle h were those that made it possible to obtain calibration models with good prediction. . While, for sparkling wines, including whites and rosés ((Group III and IV), the results show that the predictive models generated for titratable acidity, pH, density and alcohol content, can be used in practice by wineries, as they also present satisfactory parameters of cross-validation. The DESIR liquid sample concentration technique demonstrated the potential to improve the cross-validation determination coefficient (R2) and the RMSECVr of calibration models for determining phenolic compounds in red wines. In order to obtain a larger number of satisfactory predictive models for analyzing the quality of fine and sparkling wines from the São Francisco Valley's Sub-medium using a portable Vis / NIR spectrometer, it is recommended that a larger number of samples be included in the calibration set. Another alternative would be to use other techniques for data modeling.