A decision-support system to predict grape berry quality and wine potential for a Chenin vineyard
2022
Mejean Perrot, Nathalie | Tonda, Alberto | Brunetti, Ilaria | Guillemin, Hervé | Perret, Bruno | Goulet, Etienne | Guerin, Laurence | Picque, Daniel
Grape berry ripening is a complex process, and predicting the quality of wine starting from the ripening kinetics of grape berries is a challenging task. To tackle this problem, we present a decision-support system based on coupling expert know-how with probability laws encapsulated in a probabilistic model, a dynamic Bayesian network. The proposed approach predicts the ripening kinetics of grape berries starting from initial measurements and weather conditions, and then exploits the information to evaluate the potential of the wine that will produced from them. The results show that the dynamic Bayesian network predicts the total acidity concentration and the sugar content of the grape berries with a small amount of error (mean of 6% for total acidity concentration, 10% for sugar content) that is considered satisfying by the experts, making it possible to predict the ideal moment for harvesting the grapes up to two weeks in advance. Moreover, feeding the results from the probabilistic model to a fuzzy expert model, the predicted trajectories are compared to an ideal trajectory described by wine experts and formalized mathematically. From this comparison, it is possible to anticipate drifts in wine sensory quality right from the step of grape ripening.
Show more [+] Less [-]AGROVOC Keywords
Bibliographic information
This bibliographic record has been provided by National Agricultural Library