Repositori URV

Identificador: imarina:9187293

Handle: https://hdl.handle.net/20.500.11797/imarina9187293

Autores:
Schorn-García DCavaglia JGiussani BBusto OAceña LMestres MBoqué R

Resumen:
The goal of this article is to guide the reader through the critical points to be faced when monitoring a fermentation following a Process Analytical Technology (PAT) approach. To achieve this purpose Attenuated Total Reflectance – Mid-Infrared (ATR-MIR) spectroscopy coupled to chemometric techniques are proposed. Each of the crucial steps (set up of microvinifications, sampling, spectroscopic analysis and chemometric data treatment) is deeply investigated, revealing how the sampling is decisive for the subsequent modeling phase, suggesting how to set parameters to obtain good quality signals, and explaining how to prepare the data for the chemometric modeling and to perform the calculations. The modeling strategies here presented, based mainly on basic chemometric tools such as principal component analysis and partial least square regression, proved to be effective to the purposes and affordable even for non-expert chemometric users. The article shows, using real examples, how to obtain or predict several parameters from a fermentation data set – control of the fermentation evolution, prediction of oenological parameters during the alcoholic fermentation and detection of deviations from the normal operation condition.