Research On Predictive Control Of Flue Gas Denitration In Cement Kiln Based On Deep Belief Network

In recent years,China has paid more and more attention to environmental protection,and the stricter emission standard of nitrogen oxides has been promulgated for cement industry.Most cement kilns use SNCR technology to reduce the content of nitrogen oxides in flue gas.The SNCR denitration system of cement kiln has the characteristics of strong coupling and large delay.At present,most cement enterprises in China adopt the manual control or conventional PID control mode,then make the problem of too much or too little ammonia injection.The current control mode cannot meet the requirements of flue gas denitrification control.How to control the denitration process of cement kiln stably and efficiently has become a research hotspot.In this paper,the prediction model of NO_xconcentration in flue gas discharged from cement kiln is established.The control system of flue gas denitrification is designed by using generalized predictive control model.The software of predictive control of flue gas denitrification is designed and developed to realize the stable control of NO_xconcentration in flue gas.The specific research work is as follows:Firstly,the process flow of the new dry process cement production line is briefly introduced,and the mechanism of NO_xgeneration and denitrification process is analyzed in detail.The relevant factors affecting the NO_xgeneration and denitrification effect are summarized,and the application of predictive control to realize flue gas denitrification control is proposed.Secondly,aiming at the characteristics of strong coupling of variables and large detection delay of denitrification effect in cement kiln denitrification process,a prediction model of flue gas NO_xconcentration based on deep belief network is established.The data including the characteristics of denitration process are sent into the model input layer by time series method.Through the learning and mining of input data by deep belief network,the influence of strong coupling and large delay of denitration process on NO_xconcentration prediction is solved,and the accurate prediction of flue gas NO_xconcentration is realized.The validity of the time series input method is verified by experiments with the actual cement production data.Finally,based on the predictive model of flue gas NO_xconcentration based on deep belief network,the generalized predictive control model formula of cement kiln flue gas denitrification control system is deduced in detail,and the predictive control algorithm of cement kiln flue gas denitrification is obtained,and the cement kiln flue gas denitration control system using generalized predictive control model is constructed.Then,according to the predictive control algorithm of flue gas denitrification of cement kiln,the predictive control software of flue gas denitrification of cement kiln is designed and developed by using C#programming language.The simulation experiment is carried out on the predictive control software of flue gas denitrification,and shows that the designed control system has good performance.