Search

Although very often cited in publications dealing with food products, electronic noses still pose metrology problems. Lack of reproducibility is one of the more crucial problems, prohibiting any on-line quality control measurement based on an aroma reference previously taken. The database drift is due either to the intrinsic drift of the sensor or to the variations of experimental conditions. In order to quantify the influence of these conditions on the sensor performance, three screening experimental designs are performed on standard solutions, containing wine off-flavour volatiles. Three parameters are studied: the head-space temperature, the measurement cell temperature and the way of injecting samples. Their influence on indexes representing repeatability and discrimination allows the determination of the best experimental set-up.

Although very often cited in publications dealing with food products, electronic noses still pose many problems. One is the extraction of features from the response curve; in general, only the adsorption maximum is retained and input into a classification system. This paper describes a statistic-based methodology developed to extract the most pertinent features from the outputs of SnO2-gas sensor array. Several features are extracted from each sensor curve and also from its primary and secondary derivatives. They are then sorted, taking into account three specific indexes, designed to describe the repeatability, discrimination power and their redundancy. To generalise this approach, the acquisitions are carried out using various operating conditions. This protocol is applied to a set of model mixtures, representing wine with satisfactory and unsatisfactory tart or vinegar flavour. This paper shows that relevant information can be obtained from the curve maximum, but also from features related to derivatives. Moreover, the most efficient features are the same for the five sensors, which would seem to indicate that they should also be the most suitable ones for all SnO2 sensors.

A model of simulation of wineries operation has been proposed. Each elementary set of knowledge used in the model has been defined as an entity gathering constant data, variable parameters, and expert calculation functions. The model has been structured hierarchically. It has been implemented in object oriented programming. This expert simulation software produces dynamic balances of resources. This paper presents an application of the software to estimation of water consumption and organic matter rejections. At the winery Caves du Sieur d’Arques in Limoux, both predictions proved to be accurate with a maximum error of 20 % for the 10-day period of the peak of activity of grape-picking.

This paper aims at assessing the capability of high-speed analytical devices, such as aroma sensors ("electronic noses"), Fourier Transform Infra Red (FT-IR) and ultraviolet spectrometers to classify white grape musts (grape juices before fermentation) in Transform InfraRed (FT-IR) and ultraviolet spectrometers to classify white grape musts (grape juices before fermentation) in variety categories. Due to the complexity of the signal generated, specific data processing techniques have been developed and are described here. First, a pre-processing technique, based on Genetic Algorithms, is applied to spectra to improve spectrometer efficiency without expert knowledge in spectrometry ; by selecting the most discriminant subsets of wavelengths, this stochastic method tends to reduce over-fitting and improves classification results. Secondly, the Partial Least Squares Regression technique is adapted to a pattern recognition problem, using Partial Least Squares-Discriminant Analysis, a multivariate classification technique. These devises and data processing techniques are applied to ore than 100 must samples. FT-IR spectrometry is the most satisfactory technique with a 9,6% classification error level. Finally, outputs of the there individual sensors are combined in a "low-level" fusion method, by concatenating the individual sensor signals. This straightforward fusion method does not significantly improve results.2003 Elsevier Ltd. All rights reserved.; Cet article a pour objectif d'évaluer la capacité de systèmes d'analyses ultra-rapides tels que les capteurs d'arômes, la spectrométrie infrarouge à transformée de Fourier et la spectrométrie ultraviolet pour classer des moûts de raisin blanc (avant fermentation) en fonction de leur variété. Du fait de la complexité des signaux générés , des procédés spécifiques de traitement de l'information sont développés et décrits. Tout d'abord une première technique, basées sur les algorithmes génétiques est appliquée aux spectres pour améliorer l'efficacité du traitement sans faire appel à un savoir-faire d'expert en spectrométrie ; en sélectionnant le jeu de longueurs d'ondes le plus discriminant, cette technique stochastique évite le sur-apprentissage et améliore la classification. Ensuite, la régression aux moindres carrés partiels est adaptée à un problème de reconnaissance de formes : c'est l'analyse discriminante aux moindres carrés partiels, technique de classification multivariée. Ces équipements et ces techniques de traitement sont appliqués à plus de 100 échantillons de moûts. La spectrométrie FT-IR est la plus satisfaisante avec 9,6% d'erreur de classification. Les sorties de ces capteurs sont combinées entre elles par une fusion bas niveau (concaténation des signaux individuels). Cette méthode ne permet pas d'accroître la performance de manière significative.

A winery model has been investigated for dynamic simulation. The objective of the research reported here has been the implementation of uncertain enological expertise in the simulator. Uncertainty in our application is made of imprecision and preferences. Both types of uncertainty have been formalized by means of the possibility theory. Thus they become soft constraints. The enologist's expertise involved in daily tasks planning has been emphasized. This imprecise knowledge could have been integrated to heuristics miming an expert. Instead, it has been prefered to propose its integration to an algorithm of soft constraints satisfaction for calculating optimal solutions in daily tasks planning. This approach seems to suit well plant operation management as well as plant design.; Un modèle de cave vinicole a été développé pour en simuler le fonctionnement dynamique. L'objectif de la recherche présentée est l'implémentation de l'expertise oenologique incertaine dans la simulation. L'incertitude se présente sous la forme d'imprécision et de préférence. Les deux types d'incertitudes ont été formalisées par la Théorie des Possibilités. Ce sont des contraintes flexibles. L'expertise de l'oenologue a été analysée sous l' angle de la planification des tâches journalières. Cette connaissance imprécise aurait pu être intégrée à des heuristiques imitant le raisonnement de l'expert. A la place elle a été intégrée à un algorithme de satisfaction des contraintes flexibles pour calculer les solutions optimales dans la planification des tâches journalières. Cette approche semble être bien adaptée à la gestion du fonctionnement de la cave ainsi qu'à sa conception.