Prediction of Dough Rheological Properties Using Spectrum Analysis and Neural Networks

摘要

A neural network was designed to predict the rheological properties of dough from the torque developed during mixing. The back propagation neural network and GRNN network were designed and trained with the acquired mixing torque curve or its frequency domain data (input), and the measured rheological properties (outputs). The trained neural network accurately predicted the rheological properties (>96%). Prediction of dough rheological properties using frequency domain data was more accurate and used less training time. This development has significant potential to improve quality and minimize process variability during dough mixing. The ability to measure the rheology of every batch of dough enables on-line process control by automatically modifying ingredient delivery and process conditions.