Research On Optimal Control Of SCR System For Coal-Fired Power Plants Based On Neural Network Predictive Model

With the awareness of environmental protection improved the in the modernized society,the arrangement of flue gas in coal fired power plant was focused on.The nitrogen oxides(NOx),which is the major component in the gas,was strictly monitored by regulatory,and concentration of NOx in flue gas required to be controlled under the 50mg/m3,and Selective Catalytic Reduction(SCR)method is common used in power plants.In the past decade,the increasing of integration of renewable source,such as wind power and solar power,required the coal fired power plants to operate in a relative wide range of work conditions,which leads to a severe fluctuation of concentration of NOx in flue gas,and the concentration control of emission become more and more difficult by applying traditional control strategies.In order to insure the efficiency and environment-friendly of coal-fired power units in the course of operation,a SCR model with high accuracy and effective control strategy are necessary to optimize the ammonia injection.Researches based on analysis above were carried out in the aspects below.1.Based on the investigation and reasonable simplified of chemical reaction mechanism of SCR system,a model structure based on differential equation was proposed.The parameters of the model mechanism were identified by applied real operation data of power plants and genetic algorithm,and verification between mechanism model and real operation data of 1000MW power units was exhibited to prove the accuracy of the model.2.Investigated the characteristic of SCR system and control performances of traditional Control system for SCR system.3.According to a comprehensive analysis of characteristic for SCR mechanism model,an optimal control based on neural network predictive model was proposed to optimize the ammonia injection.Effectiveness of both denitration and ammonia emission were investigated in the simulations.4.For a further verification of control performance for novel control system,simulations by using real operating data in power plants were accomplished.In particular,denitration efficiency was focused and discussed.