Aplicação da espectroscopia NIR e análise multivariada na determinação de características físico-químicas e nota sensorial de café arábica

• Ano de defesa: 2018
• Autor(a) principal: Araújo, Cintia da Silva
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Federal do Espírito Santo; BR; Mestrado em Ciência e Tecnologia de Alimentos; Centro de Ciências Agrárias e Engenharias; UFES; Programa de Pós-Graduação em Ciência e Tecnologia de Alimentos
• Palavras-chave em Português: Coffee; Physicochemical composition; iPLS; Multivariate calibration; Composição físico-química; NIR; Regressão iPLS; Calibração multivariada; Café - Torrefação; Alimentos - Composição - Tabelas; Espectroscopia de infravermelho; Ciência e Tecnologia de Alimentos; 664
• Link de acesso: http://repositorio.ufes.br/handle/10/8096

Coffee is a product of worldwide interest. Its drink is one of the most consumed and appreciated in various regions of world. The physico-chemical composition of the grains is influenced by several factors and has direct interference on the final characteristics presented by the beverage. The methods of physical-chemical and sensory analysis used in the analysis of coffee samples are time-consuming and laborious, a fact that motivates the search for alternative forms of analysis, and NIR spectroscopy can be highlighted as a promising tool for this purpose. Thus, the objective of this work was to develop multivariate calibration models to determine several physicochemical and sensory properties of samples of coffee grown in several regions of the state of Espírito Santo in a fast and non-destructive manner. The coffee samples were submitted to moisture analysis, total soluble solids, pH, titratable total acidity, total and reducing sugars, potassium leaching, electrical conductivity and total phenolic compounds. Spectra no pre-treatment and preprocessed by different techniques were used for the construction of calibration models, using the method of partial least squares per interval (iPLS). The models developed showed good correlations with the values obtained by the conventional analyzes, with emphasis on the sensorial analysis, whose model obtained the highest correlation value among all models developed. On the other hand, for the analysis of humidity, the model presented significant bias, indicating that this was not adequate to estimate the moisture content of the grains. It has been found that NIR spectra can be used to determine various coffee properties and that the use of spectral preprocessing techniques has improved the ability of calibration models to estimate reference values.