Prediction And Operation Optimization Strategy Of Key Parameters Of Desulfurization System Based On Machine Learning Algorithm

Currently,most thermal power generation units use limestone gypsum wet desulfurization technology to remove sulfur dioxide from flue gas.Under the requirement of ultra-low emissions,many wet flue gas desulfurization systems have shortcomings such as large design parameter margins and poor control strategy flexibility,which could result in the problem of emission reduction but not low-carbon.Under the current trend of flexible transformation of coal-fired units,the operation control strategy of wet flue desulfurization systems have been put forward higher requirements due to frequent fluctuations in boiler load,continuous changes in flue gas load and other parameters.Therefore,research on intelligent operation and flexibility transformation of desulfurization systems should be conducted in order to achieve dual elements of low-carbon operation and environmental protection,which is of great significance to improve the operation economy of the unit desulfurization system.This article takes the limestone gypsum wet flue desulfurization system of a 660MW coalfired unit as the research object.Based on the principle of meeting the pollutant emission concentration standards and optimizing the operation economy,the history data of the desulfurization system are used to carry out modeling and optimization research,and the operation optimization strategy is given.The main research contents are as follows:Firstly,the desulfurization mechanism and process flow of a 660 MW coal-fired unit desulfurization system are introduced.The DCS system and the collected date set of the desulfurization system are illustrated,and the collected date set consists of 50 types of data indicators.The abnormal data values are processed after data pre-processing to obtain valid data samples,which provide data samples to support the establishment of data model in the next step.Secondly,the composition of the operating cost of the desulfurization system and the proportion of each cost are analyzed.The unit desulfurization cost is used as an economic indicator to analyze the impact of various parameters during operation(unit load,inlet flue gas parameters,operating parameters,outlet concentration)on the economic performance of the desulfurization system.This can provide a theoretical basis for the selection of input variables in the next prediction model.Then,a prediction model of export SO2 concentration is established.Input variables are determined through mechanism analysis and correlation analysis.BP neural network,LSTM neural network,and LSTM neural network optimized by PSO are selected to predict and model the export SO2 concentration,with RSME of 5.2936 mg/m3,2.0728 mg/m3,0.8393 mg/m3.The PSO-LSTM neural network model with the best and most prediction performance is selected for subsequent research.Finally,an operation optimization model of desulfurization system is established.The lowest unit desulfurization cost is taken as the objective function,and the predicted outlet SO2 concentration less than 35mg/m3 and the range of adjustable operating variables are taken as the constraint condition.The adjustable operating variables such as the number of slurry circulating pumps,the number of oxidation fans,the pH value of the absorption tower,and the slurry level of the absorption tower are optimized.Subsequently,the Whale Optimization Algorithm(WOA)is used to find the optimal solution set.The results show that the unit desulfurization cost of the optimized desulfurization system is reduced by 10.68%compared to before optimization.By analyzing the optimized operating variables,the optimal operation strategy was put forward in term of the recommended number of circulating pumps to be started,the recommended number of oxidation fans to be started,the optimal slurry level in the absorption tower(7.8 m～8.2 m),the optimal pH value(5.2～5.5),and the optimal combination method of circulating pumps under different SO2 mass flow rates.These suggestions can guide the optimal operation of the desulfurization system,effectively reduce the unit desulfurization cost of the desulfurization system,and improve the operational economy.