Artificial neural network model structure in the control of cane sugar crystallization

It is difficult to model through conventional methods in crystallization process of cane sugar, causing a hindrance in the automation in the industry. Based on practical production records, a concrete three layer network structure is established and test results of the artificial neural network model were compared to a regression model output. The data showed that its memory and generalization ability were higher than regression model, and its general deviation of training sample and test sample was less than that of the regression model. The artificial neural network model behaved better than the regression model in its control effect.