Premium
Artificial Neural Network Modeling of Distillers Dried Grains with Solubles (DDGS) Flowability with Varying Process and Storage Parameters
Author(s) -
Bhadra Rumela,
Muthukumarappan K.,
Rosentrater Kurt A.
Publication year - 2011
Publication title -
cereal chemistry
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.558
H-Index - 100
eISSN - 1943-3638
pISSN - 0009-0352
DOI - 10.1094/cchem-12-10-0179
Subject(s) - angle of repose , artificial neural network , chemistry , biological system , robustness (evolution) , coefficient of determination , linear regression , approximation error , mean squared error , mean squared prediction error , mathematics , process engineering , statistics , machine learning , composite material , materials science , engineering , computer science , biochemistry , gene , biology
Neural network (NN) modeling techniques were used to predict flowability behavior of distillers dried grains with solubles (DDGS) prepared with varying levels of condensed distillers solubles (10, 15, and 20%, wb), drying temperatures (100, 200, and 300°C), cooling temperatures (–12, 25, and 35°C), and storage times (0 and 1 month). Response variables were selected based on our previous research results and included aerated bulk density, Hausner ratio, angle of repose, total flowability index, and Jenike flow index. Various NN models were developed using multiple input variables in order to predict single‐response and multiple‐response variables simultaneously. The NN models were compared based on R 2 , mean square error, and coefficient of variation obtained. In order to achieve results with higher R 2 and lower error, the number of neurons in each hidden layer, the step size, the momentum learning rate, and the number of hidden layers were varied. Results indicate that for all the response variables, R 2 > 0.83 was obtained from NN modeling. Compared with our previous studies, NN modeling provided better results than either partial least squares modeling or regression modeling, indicating greater robustness in the NN models. Surface plots based on the predicted values from the NN models yielded process and storage conditions for favorable versus cohesive flow behavior for DDGS. Modeling of DDGS flowability using NN has not been done before, so this work will be a step toward the application of intelligent modeling procedures to this industrial challenge.