Research On Model Predictive Control Of Cement Kiln Flue Gas Denitration Model Based On MT-BiLSTM Algorithm

The cement industry is one of the main emission sources of industrial waste gas in China.To reduce environmental pollution,it is essential to realize the automatic control of cement denitration system and precise control of flue gas emissions.However,the characteristics of nonlinearity,delay and strong coupling in the cement production process pose challenges to the control system.Excessive pursuit of denitrification efficiency often leads to a large amount of ammonia escape,resulting in secondary pollution of the environment.To solve the above problems,this paper takes cement kiln flue gas denitrification as the research object,and uses a bidirectional long and shorttime memory loop network to build a denitration system control strategy of Multiple Indicator Model Predictive Control(MI-MPC).This strategy can achieve stable and precise control of the flue gas emission concentration of the kiln in the cement denitration system.The specific research work of this paper is as follows:Firstly,this paper analyzes the process control technology of process industry,and based on combining the current situation of model predictive control and the research on the process of cement denitration system,a corresponding scheme is proposed to solve the problems of nonlinearity,strong coupling and large delay in cement denitration system,and the relevant variables of the denitration system control strategy are determined.Secondly,a systematic model for predicting the emission concentration of kiln flue gas is proposed.Based on the bidirectional long-short-term memory loop network,the model introduces the time series layer to effectively extract the temporal characteristic information in the training data.At the same time,the random inactivation strategy is used to improve the generalization of the model and the adaptive momentum estimation method is applied to improve the prediction accuracy of the system model.Then,aiming at the problem of denitration process control,a method combining bidirectional long-term short-term memory network and model predictive control is proposed.To reduce the amount of ammonia and reduce the amount of ammonia escape,the ammonia escape amount is introduced into the objective function,and the rolling optimization solution is carried out based on the differential evolution algorithm to achieve stable control of the nitrogen oxide emission concentration and ammonia escape amount in the cement denitrification system.Finally,by simulating the data collected at the cement production site,the system model of the multi-objective time series Bi-directional Long-and Short-Time Memory network(MT-Bi LSTM)proposed in this paper is compared with the current representative statistical regression model,and the feasibility of the model is verified.The variable correlation analysis and robustness experimental verification of the control strategy proposed in this paper prove the stability of the control method,which can achieve stable and accurate control of the cement denitration system target.