Neural model of relative mass flows of water and sugar in an installation for crystallizing and centrifuging A massecuite

Three models of relative mass flows of water and sugar concerning crystallizing and centrifuging of A massecuite are presented. The input data for the calculation of particular flow volumes were the dry substance content wDS and sucrose content wS values measured in them during the sugarbeet campaign in one Polish beet sugar factory. These values, along with authors’ own knowledge and technological charts, provided the authors with the basis for the development of a mathematical model making possible the determination of appropriate, relative mass flows (sugar mass flow was accepted as the reference value) in a station for crystallizing and centrifuging of A massecuite. Then, an external balance model was developed to check the accuracy of the measured and of the calculated data. This model was provided with reliable data to develop a neural model. The developed and trained artificial neural network allowed for the simulation of relative water and sugar mass flows in an installation for crystallizing and centrifuging A massecuite. The results were satisfactory and the neural model developed was found to have a practical application in the simulation of the influence of particular sugar and water mass flows on the work of this installation.