Research For Modeling Of Flue Gas Pollutants Formation-removal Process And Intelligent Regulation Of Treatment Devices

Coal-fired power generation is currently the main power supply method in China.Pollutants such as SO₂,particulate matter,and CO₂ emitted from coal combustion are significant contributors of regional air pollution and climate change.Under the major demand of clean and efficient utilization of coal,it is of great significance to promote the ultra-low emission of coal-fired power plants and the energy saving and consumption reduction of pollutant removal system,in order to fight air pollution and to realize he goal of"carbon peak and carbon neutrality"Under the goal of"carbon peak and carbon neutrality",the large-scale integration of renewable energy sources such as wind and solar,along with the co-combustion of zero-carbon or low-carbon fuels in the furnace,have posed challenges in the context of coal-fired power generation,such as the increased fluctuations in load and frequent fuel variations.Achieving stable ultra-low emissions of pollutants and reducing system energy consumption are facing significant challenges.To address these challenges,a holistic approach based on"model development-intelligent regulation-application and validation"was followed by this paper.The Predictive models for key parameters related to the formation and removal of flue gas pollutants were established.The intelligent regulation technologies for pollutant removal processes are developed.And,an application validation studies were conducted in coal-fired power plants.Stable ultra-low emissions of pollutants are not only achieved by this research but also the energy consumption of pollutant removal systems was reduced.Support for energy conservation and cost reduction in air pollution control equipment was provided,and the automation and intelligence of equipment operation were enhanced by the result of this dissertation.The specific research contents are as follows:（1）Predictive models for key parameters in the flue gas pollutant formation and removal process were establishedIn the context of the formation-removal process of SO₂ in flue gas,the key influencing factors of SO₂ generation concentration were screened through a method that combined knowledge analysis（SO₂ generation mechanism）with data analysis（mutual information method）.A prediction model for SO₂ generation concentration was constructed based on a Long Short Time Memory（LSTM）network with autoregressive variables.The prediction of the inlet SO₂ concentration of the desulfurization device was achieved,with a Root Mean Squared Error（RMSE）of 24.6mg/m³.A hybrid modeling method driven by with dual-layer data correction was developed.The correction parameters were introduced into the mechanism/empirical formulas of SO₂ removal process,and were solved through the Particle Swarm Optimization（PSO）algorithm,and a parameter identification model was established.The LSTM network was employed to compensate for the error of the parameter identification model,and a prediction model for key parameters in the SO₂ removal process were established,enabling the prediction of the outlet SO₂ concentration of the desulfurization device with an RMSE of 0.51 mg/m³.In the context of the formation-removal process of particulate matter in flue gas,a predictive model for the formation concentration of particulate matter and other key parameters was developed based on the principles of particle generation during combustion.Accroding to the hybrid modeling method driven by with dual-layer data correction,correction parameters were introduced into the mechanism/empirical formulas of the removal process of particulate matter,and were solved through the PSO algorithm,resulting in the establishment of the parameter identification model.An Attention-Long Short Time Memory（A-LSTM）network,incorporating an attention mechanism,was utilized to compensate for the error of the parameter identification model,and a prediction model for key parameters in the particulate matter removal process were established.The model successfully achieved the prediction of the outlet particulate matter concentration of the dust removal device,with a RMSE of 1.58 mg/m³.（2）Intelligent regulation technologies for flue gas pollutant treatment devices in the flue gas were developedIn the context of the wet flue gas desulfurization device,an intelligent regulation method was established for both the absorption subsystem and the oxidation subsystem.Concerning the absorption subsystem,an investigation into the energy consumption and SO₂ removal performance of different circulation pump combinations was conducted.Addressing the optimization problem of circulation pump combination,an improved traversal algorithm based on similar operating conditions was designed.The optimal combinations of circulating pumps under various operating conditions and outlet SO₂ concentration targets were obtained.Concerning the oxidation subsystem,the influence patterns of natural oxidation and forced oxidation processes in the desulfurization device were explored.Key parameters in the control method were determined through testing the change rate of sulfite concentration in the slurry.The recommended oxidation airflows under different operating conditions were obtained.Simulation studies on a 1000MW coal-fired wet desulfurization unit indicated that after intelligent regulation,the energy consumption of circulation pump could be reduced by over 20.0%.At a unit load of 50%,the recommended oxidation airflow was around40%of the full load.In the context of the electrostatic precipitator,an intelligent regulation method was established for the high-voltage discharge subsystem and the anode rapping subsystem.Concerning the high-voltage discharge subsystem,an investigation into the energy consumption of the electric field and the particulate matter removal effectiveness under different secondary voltages was conducted.Addressing the optimization problem of the secondary voltage,constraints beyond the boundary of the optimized parameter were described using a penalty function.The PSO algorithm was employed to solve the problem and the secondary voltage under various operating conditions and outlet particulate matter concentration target settings were obtained.Concerning the anode rapping subsystem,factors influencing the ash accumulation rate and energy consumption of the dust layer were investigated.Recommended rapping intervals were obtained under constraints for ash thickness and optimal energy consumption.Simulation studies on a 300MW coal-fired electrostatic precipitator indicated that after intelligent regulation,the operational energy consumption could be reduced by 17.6%compared to empirical operation and by 42.0%compared to maximum power.For the first field of the precipitator in each chamber,rapping intervals were generally between 250-340 seconds,while for the fifth field of each chamber,rapping intervals ranged from 50,000 to 70,000 seconds.（3）An application study on intelligent regulation technology of flue gas pollutant treatment systems was conductedA four-layered intelligent regulation system for pollutant removal was established,consisting of equipment layer,edge control layer,data transmission layer,and intelligent regulation layer.For the wet flue gas desulfurization device,a real-time error feedback was employed to compensate for changes in characteristics of the absorption subsystem.The frequent switching of circulating pumps was suppressed by using a hysteresis zone,and the control of the oxidation fan was achieved through on/off regulation and frequency regulation.The industrial validation results on a1000MW coal-fired power plant’s wet flue gas desulfurization device indicate a 13%reduction in energy consumption after the implementation of intelligent regulation technology.For the electrostatic precipitator,terms for power loss,emission exceedance,emission stability loss,and terminal error were introduced into the objective function.Model errors resulting from characteristic variations were corrected through feedback.The secondary voltage was determined using a rolling optimization approach.The timing of rapping was simultaneously determined based on constraints related to ash accumulation thickness and optimal energy consumption.Industrial validation results on a 300MW coal-fired power plant’s electrostatic precipitator show a 37%reduction in energy consumption after the application of intelligent regulation technology.