Study On Combustion Optimization Adjustment Strategy Based On Real-time Measurement Of NO_x/CO And Temperature Field In Furnace

At present,in the general boiler combustion operation adjustment process,the negative feedback adjustment method based on the tail flue pollutant parameters is mainly used.However,this method has low sensitivity,poor accuracy,and serious hysteresis.The adjustment of combustion depends more on the operating experience of the operating personnel.When the combustion conditions in the furnace change,it is impossible to control the combustion in a timely manner.This paper proposes a method for predicting the temperature field and composition field in the furnace,which can quickly display the temperature field,composition field image and related values when the operating conditions parameters are changed.This method can provide certain guidance for timely adjustment of combustion.Firstly,numerical simulation was used to simulate the calculation of an actual boiler.Through the analysis of the flow field,temperature field,and component field,the practical significance of finding the components in the furnace to optimize and guide the boiler combustion was analyzed.This paper set up several sets of typical working conditions for numerical simulation,and got the data training set sum for the subsequent prediction model establishment.Relevant parameters of the actual boiler combustion were measured through field tests.For the main pollutants generated by the combustion in the furnace,the relationship between the amount of production,the generation characteristics and the combustion process was analyzed.The actual temperature field was compared with the temperature field obtained by numerical simulation.The comparison results showed that the difference between the actual temperature field value and the temperature data obtained by numerical simulation was small,and the distribution of the temperature field in the furnace could be accurately reflected by numerical simulation.And the reliability of using the numerical simulation results as a training set for the prediction model was further verified.The least-squares support vector machine was used to predict the temperature field and the component field in the furnace,so that the temperature field,the component field image and the related values could be quickly displayed when the working condition parameters were changed.The RBF kernel was selected as the model kernel function,and the optimization parameters of the kernel function were calculated.The relative errors of the prediction model for the test sample temperature,CO component and NO component prediction results compared with the numerical simulation results under the same operating parameters are 3.2%,7.4% and 3.5%,which meet the actual requirements of the project.And when the value of the component emission requirements of the furnace outlet was input,this software could provide a solution that meets the operating parameters under the condition of the emission amount.Finally,the paper visualized the predictive model through the graphical user interface in MATLAB and generated an independent EXE application that could be run on other computers without MATLAB software.