基于RBF神经网络的内模控制在热电厂脱硫系统中的应用研究

发布时间：2018-04-26 07:33

本文选题：PH值控制 + 内模控制　；参考：《青岛科技大学》2017年硕士论文

【摘要】：热电厂在运行过程中会排放大量SO2,从而造成严重的环境污染,因此对SO2的排放进行合理控制成为目前我国热电厂急需解决的重要问题之一。热电厂脱硫过程中吸收塔内浆液PH值控制过程是一个典型的非线性和大滞后系统,该控制过程具有多变量、非线性以及变增益等显著特点。本文就PH值控制的特点进行分析,提出了一种基于改进的RBF神经网络的内模控制来实现对PH值的控制。首先对吸收塔内浆液PH值控制过程进行建模,通过深入研究脱硫过程中吸收塔内浆液PH值控制的工艺过程,对PH值的特性及中和反应过程进行了分析。对于浆液PH值控制过程的数学模型运用基于Hammerstein模型最小二乘辨识方法进行辨识,得到其相应的数学模型。其次针对RBF神经网络存在的隐层单元数目和中心向量以及扩展参数难以确定的问题,利用减聚类算法来解决。最后将改进的RBF神经网络应用到内模控制中,通过改进的RBF神经网络对内模控制中的正向模型以及逆向模型进行辨识,形成完整的内模控制系统,这样不仅使内模控制的自适应能力得到极大的提高,同时还可以使内模控制的应用范围得到极大的扩展,因此形成一种先进的控制方案,并且将该方案应用到吸收塔内浆液PH值控制过程中。通过MATLAB仿真结果表明,基于改进的RBF神经网络的内模控制方案对脱硫系统中浆液PH值控制具有良好的控制性能、跟踪能力、自适应能力和鲁棒性,相对于传统PID控制具有显著优势。
[Abstract]:A large amount of so _ 2 is emitted from thermal power plants during operation, which results in serious environmental pollution. Therefore, reasonable control of SO2 emissions has become one of the most important problems that need to be solved in China's thermal power plants. The control process of PH value of slurry in absorber is a typical nonlinear and large delay system in desulfurization process of thermal power plant. The control process is characterized by multivariable, nonlinear and variable gain. In this paper, the characteristics of PH control are analyzed, and a kind of internal model control based on improved RBF neural network is proposed to realize PH control. Firstly, the control process of pH value of slurry in absorber is modeled. The process of PH control in absorber is studied deeply, and the characteristics of PH value and neutralization reaction process are analyzed. The mathematical model of the slurry PH control process is identified by using the least square identification method based on the Hammerstein model, and the corresponding mathematical model is obtained. Secondly, aiming at the problem that the number of hidden layer units, center vectors and extension parameters are difficult to determine in RBF neural network, a subtractive clustering algorithm is used to solve the problem. Finally, the improved RBF neural network is applied to the internal model control. Through the improved RBF neural network, the forward model and the reverse model in the internal model control are identified to form a complete internal model control system. In this way, not only the adaptive ability of internal model control is greatly improved, but also the scope of application of internal model control is greatly expanded, so an advanced control scheme is formed. The scheme is applied to the control process of the PH value of the slurry in the absorber. The simulation results of MATLAB show that the internal model control scheme based on improved RBF neural network has good control performance, tracking ability, adaptive ability and robustness to the PH control of slurry in desulphurization system. Compared with traditional PID control, it has significant advantages.
【学位授予单位】：青岛科技大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TM621.8;TP183

【参考文献】