Since 2014,national ministries and various regions have issued strict requirements for ultra-low emissions from coal-fired thermal power plants,and the requirements for NOx and SO2 emission reduction indicators for thermal power plants have been increased(SO2 does not exceed 35mg/Nm3,NOx does not exceed 50mg/Nm3).In the actual operation process,although the traditional circulating fluidized bed has low pollutant emission,it still needs to rely on the supporting desulfurization and denitrification process equipment,and the ultra-low emission cost of the power plant is high.Actively respond to"one plant,one test",and by combining the operation data of thermal power plants and conducting big data analysis,the cost of desulfurization and denitrification of power plants can be effectively reduced.This paper takes the 240t/h medium-sized power plant A and the 130t/h circulating fluidized bed boiler of B power plant as the research object.Carry out optimization research on desulfurization,denitrification and carbon emission reduction in power plants,mainly including:establishing a real-time calculation method of Ca/S,calculating the circulating fluidized bed through the coal feed amount and feed amount of the power plant,combined with the coal quality data and operation data in the span of a year Ca/S,it was found that the Ca/S in the furnace was greater than the design value of 2.43.The CaO content of fly ash accounts for 25%.Through the measurement of laser particle size analyzer,it was found that the proportion of limestone particle size less than 10μm accounted for 47.2%,and the escape amount was large.The particle size analysis of the fly ash found that the average diameter of the fly ash was 23.88-24.18μm;the operating conditions of different bed temperatures were analyzed,and the emission concentration of SO2 and NOx increased by 20mg/Nm3 and1.5mg/Nm3 for every 10°C increase.Adjust the desulfurization and denitrification strategy under different load conditions.Using big data analysis technology,the boiler load,furnace temperature,SNCR ammonia water content,limestone input,flue gas oxygen content,and furnace pressure difference were analyzed.The big data analysis results show that the oxygen content and load of flue gas are positively correlated with NOx generation.The nozzle of SNCR is analyzed and modified to increase the contact area between ammonia water and flue gas,which effectively improves the denitration efficiency of SNCR,greatly reduces the consumption of denitrification materials,saves more than 40%of denitrification agent,and saves ammonia water by 33.8%year-on-year.In view of the variable load operation state,while adjusting the coal volume and the primary and secondary air volume,it is also necessary to reduce the amount of limestone and SNCR ammonia water in proportion to avoid excessive input of limestone and ammonia water.Based on the above analysis,a BP neural network prediction model model for SO2 and NOx concentration was built.The SO2 model and NOx model have 47 input parameters,including steam volume,furnace temperature,etc.,the learning rate is set to 0.001,and the number of training times is set to 1000.Finally,the best training effect is achieved when the hidden layer is 20.To sum up,using big data analysis technology to fully mine power plant operation data,combine machine learning with mechanism research,and keep close to production practice.The above research results provide a theoretical basis for reducing the time-domain emission concentrations of sulfur oxides and nitrogen oxides in the circulating fluidized bed of coal-fired power plants,reducing the cost of pollutant removal in power plants,and achieving the goals of pollution reduction and carbon reduction.. |