Prediction of sensory stability of beer by infrared spectroscopy coupled with artificial intelligence

Kód výsledku v IS VaVaI

<a href="https://www.isvavai.cz/riv?ss=detail&h=RIV%2F60193697%3A_____%2F23%3AN0000053" target="_blank" >RIV/60193697:_____/23:N0000053 - isvavai.cz</a>

Výsledek na webu

—

DOI - Digital Object Identifier

—

Jazyk výsledku

angličtina

Název v původním jazyce

Prediction of sensory stability of beer by infrared spectroscopy coupled with artificial intelligence

Popis výsledku v původním jazyce

Oral presentation at international symposium 15th Trends in Brewing , KU Leuven - Technology Campus Ghent, Belgium, April 2nd - 6th, 2023. The sensory instability of beer is an undesirable and spontaneous phenomenon with a significant influence on its sensory quality resulting in the stale flavour of beer. Recent methods for evaluation of beer propensity for such staling are based on storage of beer for a few months at a lower temperature (e.g. 20 °C) or a few days at an elevated temperature (e.g. 45 °C). Although the methods provide some information regarding sensory (in)stability, their drawbacks make them useable for routine and informative evaluation with difficulty as well as for data-driven decision-making in practice. For this reason, a project focused on the evaluation of the sensory stability of beer was launched by Research Institute of Brewing and Malting in cooperation with Pilsner Urquell brewery. The project aims to develop a universal tool for quick evaluation of beer propensity for sensorial deterioration during storage (directly from fresh beer sample) by combining Fourier-transform infrared spectroscopy (FTIR) and convolutional neural network (CNN), useable for a routine quality control of beer. The FTIR acts as an analytical probe observing the chemical nature of analyzed beer in form of FTIR spectrum which is processed and evaluated by the CNN. To obtain practically usable artificial intelligence, the CNN was trained by data obtained from 163 beer brands (lagers, ales, and non-alcoholic beers) periodically brewed by collaborative breweries (more than 900 samples were obtained) and naturally aged at 20 °C for 6 and 9 months. Resulted prediction performance for an independent test set expressed by the coefficient of determination was 0.83 and the prediction uncertainty was comparable with the experimental one. Besides the sensory stability, obtained FTIR spectra were also tested for the determination of some analytical parameters of beer by CNN, e.g. alcohol, OG, IBU, FAN, polyphenols, etc., in order to make the developing tool more versatile.

Název v anglickém jazyce

Prediction of sensory stability of beer by infrared spectroscopy coupled with artificial intelligence

Popis výsledku anglicky

Oral presentation at international symposium 15th Trends in Brewing , KU Leuven - Technology Campus Ghent, Belgium, April 2nd - 6th, 2023. The sensory instability of beer is an undesirable and spontaneous phenomenon with a significant influence on its sensory quality resulting in the stale flavour of beer. Recent methods for evaluation of beer propensity for such staling are based on storage of beer for a few months at a lower temperature (e.g. 20 °C) or a few days at an elevated temperature (e.g. 45 °C). Although the methods provide some information regarding sensory (in)stability, their drawbacks make them useable for routine and informative evaluation with difficulty as well as for data-driven decision-making in practice. For this reason, a project focused on the evaluation of the sensory stability of beer was launched by Research Institute of Brewing and Malting in cooperation with Pilsner Urquell brewery. The project aims to develop a universal tool for quick evaluation of beer propensity for sensorial deterioration during storage (directly from fresh beer sample) by combining Fourier-transform infrared spectroscopy (FTIR) and convolutional neural network (CNN), useable for a routine quality control of beer. The FTIR acts as an analytical probe observing the chemical nature of analyzed beer in form of FTIR spectrum which is processed and evaluated by the CNN. To obtain practically usable artificial intelligence, the CNN was trained by data obtained from 163 beer brands (lagers, ales, and non-alcoholic beers) periodically brewed by collaborative breweries (more than 900 samples were obtained) and naturally aged at 20 °C for 6 and 9 months. Resulted prediction performance for an independent test set expressed by the coefficient of determination was 0.83 and the prediction uncertainty was comparable with the experimental one. Besides the sensory stability, obtained FTIR spectra were also tested for the determination of some analytical parameters of beer by CNN, e.g. alcohol, OG, IBU, FAN, polyphenols, etc., in order to make the developing tool more versatile.

Druh

O - Ostatní výsledky

CEP obor

—

OECD FORD obor

10406 - Analytical chemistry

Projekt

<a href="/cs/project/FW03010440" target="_blank" >FW03010440: Predikce senzorické stability piva pomocí infračervené spektroskopie a umělé inteligence</a><br>

Návaznosti

P - Projekt vyzkumu a vyvoje financovany z verejnych zdroju (s odkazem do CEP)

Rok uplatnění

2023

Kód důvěrnosti údajů

S - Úplné a pravdivé údaje o projektu nepodléhají ochraně podle zvláštních právních předpisů