Research On Modeling And Operation Optimization Of Desulfurization System Of Thermal Power Unit Based On Machine Learning

With the increasingly severe environmental problems at home and abroad,my country’s requirements for environmental protection are also increasing.At present,my country’s thermal power units are equipped with flue gas desulfurization systems,and the desulfurization system has become an important part of the thermal power unit system.With the development of science and technology and the increasing environmental protection requirements,the development of desulfurization systems for thermal power units has become increasingly mature,and many power plant desulfurization systems have achieved ultra-low emissions.However,while the desulfurization efficiency is improved,a large amount of material and energy consumption will be generated.Aiming at the two important evaluation criteria of desulfurization efficiency and economy at this stage,this paper proposes a radial basis neural network algorithm based on improved FCM clustering and an improved fast non-dominated sorting genetic algorithm NSGA-II,which are applied to the desulfurization system efficiency model.Build and run parameter optimization on.Firstly,the process flow and main system equipment parameters of the desulfurization system are introduced,the operation principle of the desulfurization system is analyzed,and then the development history of the neural network and the common radial basis neural network are introduced.In order to improve the modeling accuracy,this paper proposes a radial basis neural network algorithm based on improved FCM clustering.Firstly,the gray wolf algorithm is improved,and the improved gray wolf algorithm and FCM clustering are combined to obtain the FCM clustering analysis method based on the improved gray wolf algorithm,and then the improved FCM clustering analysis method is used to optimize the radial basis neural network.The network-related parameters are verified by the UCI dataset to verify the performance of the improved algorithm.Secondly,the influence of slurry factor,flue gas factor and other factors on desulfurization efficiency was analyzed,the main influencing factors were obtained according to principal component analysis,and the secondary influencing factors were ignored,and the desulfurization system model was established by multiple linear regression.In this paper,three criteria and weighted recursive average filtering method are used to preprocess the data to remove the gross errors and random errors in the actual data.Then,the improved radial basis neural network is applied to the establishment of the prediction model of desulfurization efficiency of thermal power units,and it is compared and analyzed with the model obtained by multiple linear regression.Finally,the calculation model of thermal power plant material consumption,energy consumption and sewage cost is analyzed.The NSGA-Ⅱ algorithm is improved from two aspects of crossover and mutation,and an adaptive NSGA-Ⅱ improved algorithm is proposed.Then,taking the desulfurization efficiency and the comprehensive cost of desulfurization operation obtained in this paper as the objective functions,the improved NSGA-II multi-objective value optimization algorithm is used to optimize the key operating parameters such as slurry density,slurry p H value,and liquid-gas ratio,and the simulation results are obtained.The Pareto frontier of the two objective functions of desulfurization efficiency and desulfurization cost under working conditions optimizes the parameters of the desulfurization system of thermal power units while ensuring that the desulfurization efficiency meets the relevant national standards,so as to achieve the purpose of saving energy consumption and cost,which provides a reference for the actual operation of the subsequent desulfurization system.