| As a big country,China’s economic development needs to promote the increase of thermal power plant power generation,at the same time,the emission of NO_x also increases,leading to serious air pollution.In order to limit NO_x emissions from thermal power plants,the state has begun to implement corresponding policies,and various large thermal power plants have taken a series of measures to NO_x emissions.Currently,one of the main De NO_x methods for flue gas emissions from thermal power plants is selective Contact Reduce(SCR),which promotes the concentration control of NO_x emissions to the atmosphere.Due to the slow reaction process of SCR system,many input/output parameters,and complex factors affecting the reaction process,NO_x emission concentration control has problems of large lag and low accuracy.In this study,the NO_x emission concentration control of SCRDe NO_x system based on IDMC algorithm was proposed.This algorithm takes a domestic 1000 MW large thermal power unit SCRDe NO_x system as the research object,uses its historical working condition data,obtains the mechanism model of SCRDe NO_x system through mechanism analysis and solution,and takes it as a system model of SCRDe NO_x system.Combined with wo A-AMLSTM neural network prediction model correction,the dynamic matrix predictive control optimized by chaotic particle swarm optimization was used to facilitate the rapid,stable and accurate control of NO_x concentration emitted by SCRDe NO_x.The main research work is as follows:Firstly,the SCRDe NO_x system mechanism model is solved and constructedFirstly,the mechanism and influencing factors of chemical reaction in SCRDe NO_x system were analyzed deeply,and a series of differential equations were obtained.The kinetic parameters were solved by second-order Runge-Kutta method,and the parameters were identified by using particle swarm optimization algorithm.Finally,the mechanism model which can accurately reflect the dynamic characteristics of SCRDe NO_x system was obtained.Second,the prediction model of NO_x emission concentration in SCR system based on WOA-AMLSTM network was constructedBased on the historical data of SCR system in power plant,the Long Short Term Memory(LSTM)network model for NO_x emission prediction of SCRDe NO_x system was established.The ground input of the model is several time series data sets that affect NO_x emission concentration greatly in the system.Whale Optimization Algorithm(WOA)is used to optimize the values in the LSTM network.The attention mechanism was used to redistribute the weight of LSTM neural network to improve the adaptive ability and accuracy of the prediction model.Finally,LSTM extracted characteristic information from time series data and trained it to obtain the NO_x emission concentration prediction model of SCRDe NO_x system with high accuracy.Thirdly,the NO_x emission concentration control algorithm of SCRDe NO_x system based on IDMC algorithm is proposedOn the basis of analyzing the mechanism characteristics of SCRDe NO_x system and the deficiencies of existing control schemes,a dynamic matrix predictive control algorithm based on WOA-AMLSTM network was proposed.The SCR mechanism model which can accurately reflect the dynamic characteristics of SCRDe NO_x system was replaced by a single transfer function as the controlled object.Woa-amlstm NO_x emission concentration model was used as a prediction model to correct the output,CPSO was used to solve the objective performance optimization function of DMC to obtain the optimal ammonia injection amount,and the optimized dynamic matrix predictive control algorithm was used to optimize the NO_x emission concentration.Fourthly,design SCR denitration monitoring platform.Relying on the CEMS database and SQL Server database provided by the power plant data center,Visual Studio.NET software is used to write and develop the SCR denitration system condition monitoring platform of the upper computer,and the platform interface display is given.Figure [51] Table [7] Reference [82]... |
PDF Full Text Request