Neurones in champagne

Abstract

Acidity is an important factor in Champagne wine quality. Predicting acidity levels is therefore an indispensable requirement when determining, which wine production process to use once the must has been supplied. Various tests which have been undertaken show that those models based on neural network programming are more accurate than either linear or quadratic function models, although they still prove to be less accurate than the more conventional analytical methods. The model generated using laboratory data from small-scale wine-making processes, has been successfully tested on data from industrial-scale processes, indicating the model's robustness. This decision-making tool could in future be used by wine-makers to optimise blending and thus improve the quality of the Champagne region's wines.