Table of Contents

Return to previous page

View Article

Machine Learning Approach to Enhancing Drying Efficiency of Hop (Humulus lupulus L.)
Pinto, M. B. C. , Ghion, R. A. C., Schmidt, F. L.

Beer is the most produced alcoholic beverage in the world with a production of 1.82 billion hectoliters in 2020, in which hops remain one of the most important raw materials. Hops drying stands as a key step to reduce the moisture right after the harvest, avoiding deterioration. Drying remains an issue due to a lack of process control, high energy demand, and consequently greenhouse gas emissions. This study intends to probe into the viability of computational modeling for time prediction to optimize the drying step of processing. For that, KNN, ANN, and Random Forest algorithms were compared with conventional empirical models according to statistical error and accuracy. From the outcomes, it was constructed a model with high accuracy R² > 0.999 using the KNN and Random Forest algorithms. It demonstrates higher accuracy in comparison with conventional mathematical models as well as a simple and more rapid time prediction. The new tool developed and tested in this study enables the reduction of drying time by a model using wider process variables. Consequently, the product quality is enhanced, and the drying footprint might be reduced by more effective energy usage.

Descriptors: kilning, machine learning, modeling prediction, artificial neural network

BrewingScience, 76 (May/June 2023), pp. 48-57