Modeling moisture sorption isotherms of corn dried distillers grains with solubles (DDGS) using artificial neural network.

摘要

Moisture sorption isotherms of corn dried distillers grains with solubles (DDGS) were modelled using artificial neural networks (ANN). Equilibrium moisture content (EMC) of DDGS, with varying chemical composition, was measured at 10, 20, 25, 30 and 40 degrees C. Samples with different chemical composition were prepared by adjusting the condensed distillers solubles (CDS) and wet distillers grains (WDG) ratio during the production process in rotary drum dryers. 2 different ANN models were tested, one with ERH and temp. only, and the other model with ERH, temp. and 5 chemical components (protein, fibre, sugars, minerals, glycerol), to predict EMC. Prediction of EMC by ANN was improved by inclusion of chemical components with low RMSE values. The R2 value was 0.99 for calibration and 0.98 for validation samples. Relative importance of chemical components in the sorption process of DDGS was also determined using ANN. Protein, fibre sugars, minerals and glycerol influenced the EMC of DDGS. The effect of protein was higher (35.26%), followed by fibre (26.12%). Results underline the importance of knowledge of chemical composition to predict the sorption behaviour of DDGS.