Dynamic Modeling of the Drying Process of Corn Grains using Neural Networks

Abstract

This study examines the model development of the drying process of corn grains as a dynamic system. The appropriate use of a dynamic model for the drying process of corn grains could lead to an effective method for optimizing the system. The optimal control strategy can be determined by predicting the future behaviors of the process using a dynamic model. In this work, the dynamic characteristic of the water loss of corn grains during dynamics treatment of temperature in the drying process was measured in a continuous manner using a precise load cell. The nonlinear autoregressive with external input (NARX) neural network is used to identify and develop a model of dynamic characteristics of the drying process of corn grains. Then for model training and validation, the dynamic responses of the rate of water loss of corn grains to drying temperature were used. A three-layered NARX neural network model consists of 1-10-1 number neurons of each layer with two times delay was successfully developed to identify and make a model such a complex system. The developed model showed the accuracy of the rate water loss of corn grains during the drying process with the mean square error (MSE), and coefficient of determination (R-squared) values are 1.88892 x 10-4 and 0.891594 consecutively.