Development of the Algorithm of Determining the State of Evaporation Station Using Neural Networks

Abstract

For the rational use of thermal resources with the help of optimal control of evaporation station at a sugar factory, it is necessary to carry out the operation control of the states of evaporation station, which is determined based on the current assessments of technological parameters such as levels and temperature in cases of a station, juice and syrup consumption, thermophysical characteristics of vapor as well as the level of its consumption by technological plants of the factory. The algorithm of determining the state of evaporation station as a control object based on intelligent methods of clustering and classification was developed. The applied method of clustering based on the Kohonen self-organizing maps allowed defining a set of possible states of the object on the basis on information hidden in time series of technological parameters of evaporation stations. The application of the method of fuzzy classification allowed determining the state of evaporation station in the current moment based on the values of current parameters of evaporation station and the obtained set of possible states of an object. The developed algorithm of determining the state of evaporation station as a control object is expedient to use in automated control systems with the purpose of operational determining the state of control object in order to make timely decisions on optimal control of evaporation station.