Research And Application Of Integrated Optimization Method For NO_x Stable Ultra-low Emission In Coal-fired Power Generation

Coal-fired power plant boiler operation is affected by many factors.On the one hand,nitrogen oxides(NO_x)produced in the boiler combustion process is an indicator to characterize the emission level of boiler pollutants.Its emission will directly affect the denitrification effect of flue gas post-treatment.On the other hand,SCR(Selective Catalytic Reduction)denitrification efficiency is used as the main indicator to measure the SCR denitrification system.SCR denitrification efficiency is too high will lead to waste of ammonia injection flow.If the SCR denitrification efficiency is too low,the NO_x emissions will not meet the national standards for ultra-low emissions of pollutants.It has a significant impact on the production efficiency and economic benefits of the SCR denitrification system and the entire generating set.Aiming at the characteristics of high-dimensional,massive and multi-parameter and multi-variable coupling of coal-fired power plant control system data.The establishment of an effective boiler combustion NO_x emission prediction model and SCR denitrification efficiency prediction model are the basis for achieving stable and ultra-low NO_x emissions.In order to solve the problems existing in the optimization control of ultra-low emission stable NO_x for coal-fired power generation.This paper proposes a set of integrated optimization methods for NO_x stabilization and ultra-low emission of coal-fired power generation.The specific research contents are as follows.(1)Aiming at the coupling characteristics of coal-fired power plant boilers with multiple parameters and multiple variables.We use Extreme Gradient Boosting(XGBoost)and correlation analysis for feature selection.The key input variables that affect the NO_x emissions of boiler combustion are selected as model inputs.Construct a prediction model of NO_x emissions in the combustion process based on Long Short-Term Memory(LSTM).(2)According to the characteristics of catalytic reduction reaction of SCR denitrification system.We use mechanism analysis for feature selection.The key input variables that affect the SCR denitrification efficiency are selected as model inputs.We use LSTM neural network to establish SCR denitrification efficiency prediction model of denitrification process.(3)The optimization of the boiler combustion system and SCR denitrification system of the thermal power unit is a multi-objective optimization problem.In order to optimize the operating cost of the system.We established the cooperative operation model of the power plant boiler SCR denitrification system on the basis of the built boiler combustion process model and denitrification process model.Then we combine the LSTM neural network model and the A3 C algorithm to propose a deep reinforcement learning model based on the LSTMA3 C algorithm.The cost control strategy of NO_x stable ultra-low emission collaborative optimization for coal-fired power plants is realized.The controllable operating parameters are optimized with the goal of the minimum operating cost of the combustion process and the denitrification process.Finally,in order to verify the effectiveness of the method proposed in this paper.Based on the actual operation data of a 1000 MW power plant unit in a power plant in Guangdong,simulation experiment analysis was conducted.Experimental results are as follows: 1)The RMSE of the combustion process model based on LSTM neural network is 10.4733.The RMSE of RBF,LSSVM and RNN models are 26.0158,19.7231 and 15.9249.It proves that the LSTM model has higher prediction accuracy and better stability than the traditional model;2)The RMSE of the denitrification process model based on LSTM neural network is 1.7191.The RMSE of RBF,LSSVM and RNN models are 3.3199,3.0713 and 2.1263.It proves that the LSTM model has higher prediction accuracy and better stability than the traditional model;3)Using the LSTM-A3 C deep reinforcement learning model proposed in this paper.Calculate the NO_x stable ultra-low emission collaborative optimization cost control strategy for coalfired power plants.The optimized power plant unit meets the NO_x stable ultra-low emission standard and the SCR denitrification efficiency remains efficient,while the operating cost can be reduced by an average of 8.07%.