A fuzzy expert system predictor for beer fermentation (FESP-BF)

Beer brewing is a complex biological process requiring efficient operation and management to provide the consumer with a consistently high quality beer. The control of many quantitative and qualitative brewing parameters requires the skill of an expert brewer.; An fuzzy expert system predictor is developed which predicts the length of time required for lager fermentations based on upstream inputs. An initial prediction of fermentation time is performed using expert knowledge based on the fermenter inputs of yeast concentration and yeast volume. An update to the initial fermentation time prediction is made by predicting the diacetyl (VDK) concentration based on current measurements of pH and specific gravity. Fuzzy logic provides the expert system with a schema for uncertain reasoning, allowing system uncertainty to be incorporated in the expert system rules.; The fuzzy expert prediction rule algorithms were written in the CLIPS development language. External fuzzy logic functions for fuzzification and defuzzification were written using C and called as external functions within the expert rules.; FESP-BF predicted the fermentation time to within 24 hours of actual fermentation time for 22 out of 28 predictions. Predictions based on work VDK concentrations improved the initial predictions based on yeast concentration and yeast volumes when initial predictions were not within 24 hours of actual fermentation time. This level of prediction was considered useful by the domain expert. FESP-BF provides the brewery manager with information which could be used in both production scheduling, and management of brewery operations. Fuzzy logic was considered an appropriate tool for managing the uncertainty inherent in the qualitative aspects of lager fermentation.