Research On Predictive Control System For Boiler Nitrogen Oxides Emissions Power Station

Nitrogen oxide emissions(NOx)from coal-fired power plants are one of the important sources of environmental pollution.As environmental problems become more serious,reducing pollutant emissions has become a major task for power plants.Because boiler nitrogen oxide emissions are related to a variety of production parameters and have strong non-linear characteristics,it is difficult to effectively control nitrogen oxide emissions from coal-fired power plants.Aiming at the correlation of the modeling data set,the accuracy of the prediction model and the accuracy of the optimization control during the NOx emission control process of coal-fired power plants,the 1000MW boiler is taken as the research object,from data preprocessing,modeling methods,predictive control and system experimental research was carried out in four aspects of design.The specific research contents are as follows:Firstly,for the relevance of the modeling data set,and in order to reflect the impact of different characteristic variables on nitrogen oxide emissions,the modeling data set of the boiler production process is divided into control variables and state variables based on the variable characteristics.Lasso algorithm was used to analyze the correlation between control variables and state variables,and the control variables and state variables with significant correlation were selected as the input variables of control variable prediction model and state variable prediction model,respectively.Secondly,in view of the accuracy of the prediction model of important parameters of power plants,this paper proposes a nonlinear combination prediction method based on deep belief networks.Based on the unit load,the production data under different operating conditions are distinguished,and two different operating conditions are divided.For power plant boilers under different operating conditions,a deep confidence network is used to establish a control variable prediction model and a state variable prediction model,and then the two prediction models are combined using a non-linear method.Finally,the differences based on the non-linear combined deep confidence network(NCDBN)are established.Prediction model of important boiler parameters under operating conditions.Then,according to different working conditions,set boundary constraints of adjustable parameters such as secondary air door opening and burn-out air.Taking into account the safe and efficient operation of the unit,set unit load constraints and boiler combustion efficiency constraints.Under the conditions of boundary constraints,load constraints and efficiency constraints,a rolling optimization model with the lowest nitrogen oxide emissions as the target is established,and the particle swarm optimization(PSO)algorithm is used to solve the optimization model,and the optimal control variables are finally obtained.When the absolute errors of the NOx emissions,unit load and boiler combustion efficiency prediction model are greater than 2%for three consecutive times,the latest production data will be used to re-establish the prediction model of the main production parameters of the boiler.Finally,the C#software was used to develop a NOx emission prediction control system for power plant boilers.The predictive control system includes five major functional modules:user login,data loading,intelligent model,predictive control,and result display.The system can log in to the system interface according to the specific permissions of authorized users,and perform intelligent modeling,optimization control,parameter adjustment or result query in the interface.This system power plant operator is used for reference.The experimental results show that the proposed nonlinear combined prediction method not only accurately predicts important boiler parameters,but also lays the foundation for the predictive control of subsequent nitrogen oxide emissions.The generalized predictive control method proposed in this paper can effectively reduce the boiler’s nitrogen oxide emissions,and the prediction control error is less than 3%.