Research And Applications On Method Of Modeling For Wet Flue Gas Desulfurization System Of Thermal Power Plants

Electricity is an indispensable energy source in the developing society nowadays.It is very important to strictly control the pollutants emission such as SO₂ in the current situation of the development of thermal power industry.Limestone/gypsum wet flue gas desulfurization technology is a kind of mature desulfurization technology and be widely used currently.Wet Flue Gas Desulfurization System is also an important part of a thermal power generating unit.Wet flue gas desulfurization system is a complex multi-parameter,multivariable and nonlinear system.It is of great significance to identify the model of wet flue gas desulfurization system by using appropriate methods for analyzing the desulfurization system and controlling desulfurization efficiency.In this paper,a limestone/gypsum wet flue gas desulfurization system of300MW thermal power unit in a coal gangue power plant was the research object.Based on the in-depth analysis of the technological process and reaction mechanism of the desulfurization system,the main influential factors affecting the desulfurization efficiency were determined.Real-time measurement data of the operation was collected and preprocessed based on choosing appropriate input and output signals of the system.On the Matlab platform,wet desulfurization system in different working conditions was modeled respectively by using support vector machine?SVM?and transfer function based on particle swarm optimization?PSO?.According to the analysis of error index of each model,modeling methods and model structure were confirmed being feasible and correct.After determining the modeling methods were feasible,the two modeling methods were applied and optimized respectively.Aiming at SVM,this paper introduced the method by using ant colony optimization algorithm to optimize parameters of SVM.And used the optimized support vector machine to model the desulfurization system.Combining the results of un-optimized system's actual data output curve,the optimization of ant colony algorithm helped to improve the accuracy of SVM model.In order to optimize the model identification based on particle swarm optimization,an improved iterative formula is introduced to optimize particle swarm optimization?PSO?.After analyzing the process of particle swarm optimization and the output of new recognition model,the results verified that optimized algorithm accelerates the optimizing process and improves the recognition efficiency and identification accuracy.