Research On Multi-objective Operation Optimization Of NO_x Emission Concentration Of 1000MW Boiler

The nitrogen oxides（NO_x）of coal-fired power plants cause environmental pollution and endanger human health.Therefore,the national environmental protection department has formulated stricter NO_x emission standards.This paper took a 1000 MW coal-fired boiler as the research object,processed the abnormal values in the original modeling data,calculated the delay time,and designed an adaptive feature selection algorithm.At last,this paper established the dynamic prediction model of NO_x emission and boiler thermal efficiency.Based on a multi-objective optimization algorithm,this paper developed NO_x emission concentration optimization control system.The specific research contents are as follows:（1）Aiming at solving the problem of abnormal value caused by sensor failure and signal interference,the actual operation data are preprocessed.Use the Pauta criterion to identify the abnormal data,calculate the average value of 10 samples before the time point of abnormal value and use the average value to correct the abnormal value.Normalize the original data and eliminate the dimensional difference of the modeling data.（2）Aiming at the solving problem of establishing an accurate NO_x emissions prediction model,a dynamic prediction modeling method based on data-driven and error correction is proposed.According to the unit load,the modeling dataset is divided into three conditions: ascending condition,descending condition and steady-state condition.The NO_x measurement and combustion reaction have delay characteristics.The maximum information coefficient（MIC）is used to calculate the delay time of each input variable.And the Lasso-relief F adaptive feature selection algorithm is developed to obtain the feature set.The initial NO_x emission prediction model is established based on the extreme learning machine（ELM）.The initial prediction errors are corrected.（3）Aiming at solving the multi-objective optimization problem of reducing NO_x emissions and improving operation efficiency,the secondary air volume is selected as the control variable.One objective function is the minimum difference between the predicted NO_x emission and the expected value.The other objective function is the minimum difference between the predicted boiler efficiency and the expected value.The Pareto optimal solution set is obtained by the NSGA-II algorithm,and the optimized boiler combustion condition is analyzed.Provide operation suggestions for on-site personnel.（4）A data-driven NO_x emission concentration optimization control system is developed by using c# software.The optimization control system includes five functional modules: user login,data loading,modeling and prediction,multi-objective optimization and result display.After logging into the system,the user can query the data of the required time period through the data loading interface,and observe the changes of various parameters through the drawing display;In the modeling and prediction interface,you can select an appropriate modeling algorithm and set super parameters for modeling and prediction;The multi-objective optimization operation interface based on NSGA-II algorithm provides a variety of optimization suggestions for reducing NO_x emission concentration and improving boiler efficiency,and provides reference for field staff.Based on the actual production data,the results show that the prediction errors of the proposed method are less than 2% under different working conditions,and can accurately predict the NO_x concentration at the combustion outlet;The proposed multi-objective optimization method can meet the needs of field workers for boiler NO_x emission and boiler combustion efficiency.