VIEW ARTICLE    DOI: 10.1094/ASBCJ-52-0015

Neural Network Modeling for Predicting Brewing Fermentations. Mei-Jywan Syu and George T. Tsao, Laboratory of Renewable Resources Engineering, Purdue University, West Lafayette, IN 47907-1295; and Glen D. Austin, Guy Celotto, and Tony D'Amore, Brewing Research Department, Labatt Breweries of Canada, 150 Simcoe Street, London, Ontario, Canada N6A 4M3. J. Am. Soc. Brew. Chem. 52:0015, 1994.

Brewing fermentation performance is dependent on a large array of initial ingredient and process variables. It is virtually impossible to account for every possible effect that each variable has on the fermentation profile of brewer's wort and yeast. However, recent developments in the area of neural network modeling hold application in characterizing and predicting fermentation performance. Neural networks are capable of processing large amounts of data in a parallel manner such that relationships between system inputs and outputs are identified. Through processing many sets of actual pilot and plant scale data, empirical relationships are learned by the network and, given new sets of data, predictions of fermentation performance are possible.

Keywords: Backpropagation, Ethanol, Fermentation, Neural network, Prediction