The Study Of NOx Generation Prediction Model And Combustion Optimiization For Coal-fired Power Plant Boilers

The combustion of coal and NOx generation in the boiler of a coal-fired power station boiler is a very complicated process.It is not only affected by external factors such as the type of coal and the type of the furnace,but also by its combustion mode and the air distribution mode of each secondary air.The influence of external factors such as external factors,and the interaction and coupling between various internal and external factors.So far,the academic community has not fully explored its clear,accurate and reliable mathematical models,but with the continuous improvement of statistical theory and computer labor With the rapid development of intelligent neural network technology,artificial intelligence technology has made some achievements in the prediction of NOx generation concentration and thermal efficiency of coal-fired boilers,and multi-objective optimization of boiler combustion.In this paper,a variety of artificial intelligence-based NOx generation concentration and boiler thermal efficiency prediction models are established,and different multi-objective optimization algorithms are introduced to optimize the combustion of coal-fired boilers.First,taking a 660MW boiler in a coal-fired power station as the research object,based on MATLAB2017b platform software,a radial basis neural network(RBF)and a support vector machine(SVM)boiler combustion characteristic prediction models were established respectively.17 input parameters,such as air volume,secondary air damper opening,burn-out damper opening,and furnace outlet oxygen,were used as input variables of the model.The predicted NOx concentration and boiler thermal efficiency were used as two output parameters.The prediction model shows that the maximum error of NOx generation concentration of the improved RBF neural network prediction model is 4.39%,the average error is 1.82%,the maximum relative error of the boiler thermal efficiency prediction result is 0.14%,and the average relative error is 0.06%;The maximum error of the simulation result of the NOx generation concentration of the SVM prediction model is 3.26%,the average error is 1.17%,the maximum relative error of the prediction result of the boiler thermal efficiency is 0.13%,and the average relative error is 0.05%.It shows that these two modeling methods have good accuracy and generalization ability.By quantifying and comparing the simulation and prediction results of the two models,the support vector machine method has stronger model prediction ability than the RBF neural network method.And generalization capabilities.In addition,in order to analyze the influence of different coal mill combination methods and secondary air distribution methods on the combustion characteristics of boilers in power plants,ABCDE,ABDEF and BCDEF three coal mill combination methods were selected for the combination of mills to determine the combustion characteristics of the boiler Impact analysis,and the combination of the above three coal mills and different secondary air distribution methods(equal air distribution,drum waist air distribution,inverted pagoda air distribution,and positive pagoda air distribution),The results show that the combustion concentration of NOx generated by the boiler in the ABCDE mill combination operation mode is at least 125mg/m3,while the NOx generation concentration of the boiler in the ABDEF mill combination mode is the highest;and for the boiler thermal efficiency,different secondary air distribution methods affect the boiler.The thermal efficiency has a great influence.The thermal efficiency of the boiler in the equal air distribution mode is the highest thermal efficiency of all air distribution modes to be 94.9%;and the thermal efficiency of the boiler in the regular tower air distribution mode is the lowest thermal efficiency of all the air distribution modes.Based on the RBF neural network prediction model established and completed above as the basis,the prediction result of boiler NOx concentration and boiler thermal efficiency is used as the objective function of the multi-objective optimization algorithm,and the MOEA/D and NSGA-2 intelligent optimization algorithms are used for coal combustion.Boiler multi-objective combustion optimization.The results show that compared with the MOEA/D algorithm,the pareto front obtained by the NSGA-2 optimization algorithm has better convergence,distribution,and running speed.After the optimization of the NSGA-2 algorithm,the NOx generation concentration range of the boiler is 85mg/m3 to 135mg/m3,which is 35.3%lower than the lowest generation concentration of the boiler in historical operation,and the thermal efficiency range of the boiler is 94.6%-95.2%,which is the highest compared to the historical operation of the boiler Increase efficiency by 0.6 percentage points.The main operation data of power plant boilers in the optimization results have certain reference significance for power plants to improve operation efficiency,increase economic benefits,and reduce environmental pollution operations.