APPLYING OF NEURAL NETWORK ON THE WINE SENSORIAL ANALYSIS FROM BARBADOS CHERRY

摘要: This paper presents an approach of artificial neural networks to predict the sensorial qualities wines. A Kohonen network has been used as a software tool in order increase human skills this kind application. An initial prototype was implemented using technology, together with Visual Basic from Microsoft, for evaluation seven samples Barbados cherry (Malpighia glabra L.) wine. Fifty consumers chosen, perhaps, had obtain data hedonic scale 1–9 times. Sensorial values flavor, aroma and appearance obtained dating were compared. The characteristics wine, such flavor color, similar fresh fruit. evaluated wines very good; all more than 5 scale. Results showed that classified distinct group, frequency among their responses. results or better statistical classification; shows use algorithm analysis is valid. Kohonenalgorithm good clustering wine samples, its uses analyses promising. PRACTICAL APPLICATIONS The work have practical fruit production sample selections by analysis. Thus, it optimal conditions produce effects pulp mass sugar content on evaluated. visual basic program elaborated, work, algorithm, which will be powerful taking decision beverages, foods too much products methods make comparison between them.