DEVELOPMENT OF AN ARTIFICIAL NEURAL NETWORK FOR THE MEAD PRODUCTION WITH ADDITION OF JABUTICABA PEEL EXTRACT AT HIGH CELL CONCENTRATIONS

• Grant number: 24/06851-7
• Support Opportunities: Scholarships in Brazil - Scientific Initiation
• Start date: August 01, 2024
• End date: July 31, 2025
• Field of knowledge: Engineering - Chemical Engineering
• Principal Investigator: Rafael Ramos de Andrade
• Grantee: Voluspa Michelle Jarpa Ramos
• Host Institution: Instituto de Ciências Ambientais, Químicas e Farmacêuticas (ICAQF). Universidade Federal de São Paulo (UNIFESP). Campus Diadema. Diadema , SP, Brazil

Abstract

Mead is a fermented alcoholic beverage based on water and honey produced by the action of yeasts, normally strains of Saccharomyces cerevisiae, on sugars such as glucose and fructose. The producer often faces obstacles arising from the lack of control over important process parameters. Bearing this in mind, the present work aims to develop a virtual sensor based on Artificial Neural Networks that is capable of predicting the concentrations of cells (X), substrate (S), ethanol (P) during the fermentation for the manufacture of mead with the addition of jabuticaba peel extract. To make this possible, the pH of the medium, the concentration of soluble solids (°Brix) and the optical density (O.D.) were chosen as input variables. To obtain and simulate the feedforward network with supervised training, experimental data from five fermentations will be used. The neural networks will be tested in different configurations and the best training algorithms will be identified (Levenberg-Marquardt, Levenberg-Marquardt with Bayesian regularization and Powell), activation functions, number of intermediate layers and number of neurons in each layer, in order to optimize the prediction of output variables by the network. It is expected that the virtual sensor can be used to monitor and optimize the beverage production, being able to identify the variables of interest, and enabling action on the process when necessary, to improve the performance of the fermentation and the quality of the final product. The neural network can be applied to other fermented drinks, mead variations and process conditions, as long as the network is retrained.