Clasificación automatizada de rones Havana Club y Santero mediante la espectrofotometría ultravioleta-visible y la quimiometría
Fecha
2014-06-25
Autores
Domínguez Rodríguez, Andy G.
Título de la revista
ISSN de la revista
Título del volumen
Editor
Universidad Central "Marta Abreu" de las Villas
Resumen
Esta tesis constituye un primer paso en posde lograr la clasificación automática de rones según su marca y tipicidad. Para ello, se desarrollaron y validaron modelos quimiométricos para diferenciarentre las marcas de rones cubanos Havana Club y Santero así como entre sus tipicidades a partir de los espectros UV-Vis.
Se estableció que la modelación de rones oscuros y claros se debe realizar de forma independiente para una mejor discriminación y a una relación de dilución muestra: agua destilada de 180 μL: 2 mL y 1 mL: 2 mL (respectivamente)para el registro de los espectros. Paralelamente se realizó un estudio cualitativo mediante HPLC-DAD comprobándose que los compuestos incorporados durante el añejamiento-acabado del ron determinan las propiedades de absorción UV-Vis.
Para el análisis exploratorio de los datos experimentales se aplicó el PCA y el HCA y se hizo una selección de variables mediante sus poderes de discriminación y modelación estableciéndose idóneo el rango de 215-450 nm para la lectura de los espectros. Se aplicaron los algoritmos de clasificación k-NN, SIMCA y SVM, obteniéndose los mejores modelos con el SVM que mostró 100% de sensibilidad en ambos modelos y 100 y 90.91% de selectividad para oscuros y claros respectivamente.
This thesis is a first step towards automatic classification of rums by brand and typicality. To accomplish this, chemometrics models to differentiate between brands of Cuban rums Havana Club and Santero and between their typicality from the UV-Vis spectra models was developed and validated. It was established that the modeling of light and dark rums should be performed independently for better discrimination and a dilution ratio sample: distilled water of 180 μL: 1 mL and 2 mL: 2 mL (respectively) for measuringspectra. In parallel, a qualitative study was performed by HPLC-DAD proving that incorporated compounds during aging rum-finished determine the properties of UV-Vis absorption. In the exploratory analysis of the experimental data was applied PCA and HCA, and a selection of variables was made by its discrimination and modeling powers. The range of 215-450 nm was established for the reading of the UV-Vis spectra. Classification algorithms k-NN, SVM and SIMCA were applied, yielding the best models with SVM that showed 100% sensitivity in both models and 100 and 90.91% selectivity to light and dark respectively.
This thesis is a first step towards automatic classification of rums by brand and typicality. To accomplish this, chemometrics models to differentiate between brands of Cuban rums Havana Club and Santero and between their typicality from the UV-Vis spectra models was developed and validated. It was established that the modeling of light and dark rums should be performed independently for better discrimination and a dilution ratio sample: distilled water of 180 μL: 1 mL and 2 mL: 2 mL (respectively) for measuringspectra. In parallel, a qualitative study was performed by HPLC-DAD proving that incorporated compounds during aging rum-finished determine the properties of UV-Vis absorption. In the exploratory analysis of the experimental data was applied PCA and HCA, and a selection of variables was made by its discrimination and modeling powers. The range of 215-450 nm was established for the reading of the UV-Vis spectra. Classification algorithms k-NN, SVM and SIMCA were applied, yielding the best models with SVM that showed 100% sensitivity in both models and 100 and 90.91% selectivity to light and dark respectively.
Descripción
Palabras clave
Ron, Modelos Quimiométricos Validación, Chemometrics Models, Rum, Validación, Química Anallítica, Habana Club, Havana Club