Research On Neural Network Predictive Control Of Decomposing Furnace Outlet Temperature

The new dry cement pre-decomposition production technology has gradually become the mainstream method of current cement production technology.The decomposition furnace is an important equipment in the cement production process,which undertakes about 90% of the decomposition task of cement clinker.Whether the decomposition furnace can work stably has an important impact on the output and quality of cement.This thesis first discusses the significance of cement production to China’s economic construction and development.Starting from the current research status,it briefly introduces the methods used by different scholars in the research of decomposition furnace modeling and control.Aiming at the pre-decomposition technology,the working mechanism of suspension preheater and decomposition furnace is briefly introduced,and the main factors affecting the outlet temperature are discussed.In the modeling of decomposition furnace outlet temperature,most of the traditional research methods use empirical method to select air(tertiary air),coal(coal feeding amount),and feed(raw meal)as research variables,which are difficult to reflect the internal laws of the decomposition furnace.In this thesis,using the measured data of production site,the ElasticNet can effectively screen variables without over compression,reduce the dimension of the data,eliminate the influence of irrelevant factors on the prediction model,and establish the outlet temperature prediction model based on ElasticNet and Long Short Term Memory(LSTM)neural network,The impact of manual configuration of network parameters on model error is discussed,and comparative experiments are established to verify the effectiveness of the model.Aiming at the disadvantage that the network parameters need to be manually configured and verified,an Improved Particle Swarm Optimization(IPSO)based LSTM neural network is proposed.The parameters of LSTM are optimized by IPSO to realize the automatic configuration and verification of network parameters.In the control of decomposition furnace outlet temperature,the neural network predictive control model is established by using LSTM,and the deep learning network is introduced into the predictive control.A rolling optimization controller is designed with the coal feeding rate as the control variable and the genetic algorithm as the optimization strategy.The simulation results show that the control model can effectively control the decomposition furnace outlet temperature.In the control of decomposition furnace outlet temperature,the predictive control model of cement decomposition furnace outlet temperature is established.The neural network predictive control of outlet temperature is realized by using LSTM neural network combined with the principle of predictive control,and the deep learning network is introduced into the predictive control.A rolling optimization controller based on genetic algorithm is used to design a control model with coal feeding rate as control variable.The simulation results show that the control model can effectively control the decomposition furnace outlet temperature.