| China’s energy structure determines that coal-fired power generation will still dominate for a long time,and pollutants such as particulate matter(PM),sulfur oxides(SOx),and nitrogen oxides(NOx)emitted in the process of large-scale coal-fired power generation are the main causes of the regional haze.In recent years,driven by the country’s major needs to improve the quality of the regional atmospheric environment and promote the clean and efficient utilization of coal,important progress has been made in coal-fired power generation pollutant emission reduction technology,especially the popularization and application of ultra-low emission technology for coal-fired power plants.The emission concentration of major pollutants in coal-fired power generation flue gas has been significantly reduced.However,under the background of carbon peaking and carbon neutrality,coal-fired power generating units are affected by various factors such as deep peak regulation,unstable coal quality,energy saving and consumption reduction requirements,and the operating conditions are very complex,which poses challenges to the operational reliability,stability and economy of each key device in order to ensuring the emission concentration.Based on the high reliability,stability and economic requirements of ultra-low emission systems in coal-fired power plants,this thesis discusses the construction and application of ultra-low emission intelligent control systems for coal-fired power plants.What’s more,researches on intelligent control systems development,gaseous pollutant concentrations forecast and optimization control of environmental protection systems etc.are carried out,which proposed solutions and carried out industrial verification for improving the performance of key systems such as Selective Catalytic Reduction(SCR)and wet flue gas desulfurization(WFGD)and reducing operating energy consumption and material consumption.First of all,combined the ultra-low emission system in coal-fired power plants,a“cloud-edge-device”collaborative ultra-low emission intelligent regulation system platform was designed and developed.Based on the requirements of data collaboration,model collaboration,and application collaboration,the system architecture is designed for the cloud side,edge side and terminal side,achieving the fusion processing of heterogeneous data.A distributed time series database based on hadoop and a visual modeling tool was designed and developed in order to achieve efficient model development and management.The data communication module,data storage module and data calculation module are designed and developed on the edge side to realize data collaboration,model collaboration and application collaboration between the cloud side and the edge side,and realize high-fidelity and high-throughput data transmission between the cloud side and the terminal side.It makes the cloud side and the device side better integrate.On the terminal side,based on DCS,CEMS and other systems,through the development of data upload interface and command release interface,real-time online detection and optimized operation of ultra-low emission physical equipment can be realized.Secondly,a multi-model prediction method of gaseous pollutant concentration based on LSTM neural network training is proposed for large-scale coal-fired units in order to solve the lag and distortion during the detection of flue gas pollutant concentration under the condition of large operating conditions change gradient and high switching frequency.A method based on random forest method,K-L divergence calculation,etc.and analyze the correlation between the parameters.The prediction models of nitrogen oxides and sulfur oxides are developed.The RMSE of nitrogen oxides is within 7.3mg/m~3,and the RMSE of sulfur oxides inlet is within 20mg/m~3.Thirdly,for the denitrification system,a control model identification method for the denitration system based on IPSO was proposed.Tthe influence of key parameters such as ammonia injection amount and inlet NOx concentration on the outlet NOx concentration was obtained using historical data.An accurate control model of the denitrification device was obtained.What’s more,the DMC-PID ammonia injection cascade control logic was established,which significantly improved the control level of the denitrification device.Fourthly,for the desulfurization system,the correction mechanism model of the desulfurization absorption subsystem was constructed.The operation optimization method of the desulfurization absorption subsystem was developed.The optimal configuration method of the circulating pump under different working conditions was obtained.The slurry oxidation mechanism model of the desulfurization absorption tower was constructed.The effect of reaction enhancement factors,slurry level,droplet size,oxygen content of the original flue gas and other key parameters on the oxidation process was obtained.The real-time optimization of the oxidation air volume are realized.Finally,the industrial application verification was carried out on a 1000MW ultra-low emission coal-fired unit.The cloud computing layer of the intelligent control system deploys functional modules such as equipment information management and visual modeling tools.The edge computing layer deploys the gaseous pollutant soft measurement prediction model and the optimal control model of each system,and uses the real-time data of the equipment for real-time calculation;At the terminal device layer,the optimized control results of the edge computing layer are transmitted to the DCS system in real time through the OPC interface for real-time control.The results shows that the average NOx concentration at the outlet was controlled from 35.8 mg/m~3 to 40 mg/m~3,and the average ammonia escape was reduced from 0.95 ppm to 0.74 ppm.The energy consumption of oxidation fan and slurry circulating pump decreased by 23.7%and 34.1%respectively.The industrial application verification shows that the ultra-low emission intelligent control system can effectively improve the reliability,stability and economy of ultra-low emission system. |
PDF Full Text Request