Optimization Technology Research Of Wet Desulfurization System Based On Machine Learning

At present,thermal power generation still occupies the main position of China’s electric energy,and SO₂ is one of the main air pollutants emitted by thermal power units.In order to remove the SO₂ produced by thermal power plants,most power plants use limestone-gypsum wet desulfurization systems.Aiming at the problems of system stability,economy and desulfurization efficiency in the limestone-gypsum wet flue gas desulfurization system,this paper researched on the desulfurization system characteristics,the prediction of the number of desulfurization tower circulating pumps and the optimal target value of the control operating parameters of the desulfurization system based on clustering.The number of circulating pumps in the limestone-gypsum wet flue gas desulfurization system directly affects the desulfurization efficiency of the desulfurization system.This paper used an improved integrated learning algorithm based on sampling to study the historical operating data of a 1000 MW thermal power unit under the condition of sample imbalance.And this paper used box plots to identify and delete abnormal values,and then used R detection method to filter steady-state operating condition data,finally used the integrated learning method based on clustering and sampling to balance and model the desulfurization data,and predicted the number of circulating pumps.The pros and cons of the sampled data were judged by the prediction accuracy and recall rate.The prediction accuracy of the pump combination as the model output was also discussed.At last,the established model was used to study the stable operation of the desulfurization system.In view of the complex internal reactions of the desulfurization system,the test methods are often difficult to be directly applied to practical engineering problems due to many factors.This paper used the random forest modeling method to obtain relevant characteristic models,and separately studied the influence of p H and liquid-gas ratio on the desulfurization efficiency.This model can better reflect the influence of various parameters in the desulfurization tower on the desulfurization efficiency.The prediction model was used to optimize the operation of the desulfurization system.At the same time,the total cost was taken as the objective function,and the genetic algorithm was used to optimize the main operating parameters.In addition to the number of circulating pumps,there are many factors that affect the desulfurization efficiency and desulfurization cost of the system,such as the liquid-to-gas ratio,slurry p H,and slurry density.This paper used a clustering algorithm to complete the data cluster analysis,taked the desulfurization system of a 1000 MW unit as the research object,used the algorithm to mine the target value of the controllable operating parameters of the desulfurization system,established a target library for the optimized operation of the desulfurization system.Finally a guiding example for the optimization of desulfurization system operation was given.