SCR脱硝过程神经网络建模及控制

发布时间：2018-04-23 10:01

本文选题：神经网络 + SCR脱硝　；参考：《华北电力大学》2017年硕士论文

【摘要】：随着社会的发展,环境面临着严峻的问题,大气污染作为环境问题的一部分也越来越受关注,燃煤发电作为我国的主要电力来源,这在短期内是无法改变的事实。因为氮氧化物是造成大气污染的主要因素之一,所以国家对氮氧化物排放量的标准也更加严厉。因此,如何经济有效的降低燃煤机组氮氧化物排放已经刻不容缓。SCR脱硝技术是目前国内燃煤机组应用最为广泛的脱硝方法,因此,对SCR脱硝过程进行建模及控制具有一定的研究意义。SCR脱硝过程是一个大滞后、大惯性、非线性且容易受很多因素扰动的复杂过程,因此采用传统的建模方法对其建模是很困难的。基于BP神经网络具有的高度自学习自适应能力,适用于非线性系统以及具有一定的容错能力,这些特点使其特别适合求解内部机制复杂的问题。因此本文采用BP神经网络建模方法对SCR脱硝过程进行建模,然后根据建立好的模型,采用神经网络自适应控制来实现对SCR出口氮氧化物浓度的控制。本文首先介绍了课题的研究背景及意义还有神经网络的结构、算法及建模步骤,然后对SCR脱硝过程进行机理分析,分析影响出口烟气浓度的影响因素。确定输入输出变量,根据采集某燃煤电厂600MW机组所配备的SCR脱硝系统的现场数据为基础,先对数据进行选取和预处理,再选取合适的神经网络模型结构和参数,对模型进行训练。然后采集一部分现场数据对模型的准确性进行验证,结果证明误差在可接受的范围内,这时基于SCR脱硝过程的BP神经网络模型建立完成。所建立的BP神经网络模型能正确的反映脱硝系统的动态特性。最后把建立好的BP神经网络模型作为被控对象,采用了神经网络模型参考自适应控制的方法,来实现对SCR脱硝过程出口NOx浓度的控制。利用BP神经网络作为控制器,通过不断修改神经网络的权值来减小被控对象输出与参考模型输出之间的差值,使被控对象的输出能快速的响应参考模型的输出,最终使得控制达到理想的要求。仿真结果表明,出口NOx浓度具有良好的响应特性,说明这种控制方案是可行的,为其在实际的应用打下了基础。
[Abstract]:With the development of society, the environment is faced with severe problems, air pollution as a part of the environmental problem is more and more concerned, coal-fired power generation as the main source of power in China, this is a fact that can not be changed in the short term. Because nitrogen oxides are one of the main causes of air pollution, national standards for nitrogen oxides emissions are more stringent. Therefore, how to reduce NOx emission of coal-fired units economically and effectively is the most widely used denitrification method in China. The modeling and control of SCR denitrification process has certain significance. SCR denitrification process is a complex process with large lag, large inertia, nonlinear and easily disturbed by many factors. Therefore, it is very difficult to use the traditional modeling method to model it. Based on the high self-learning adaptive ability of BP neural network, it is suitable for nonlinear systems and has certain fault-tolerant ability, which makes it especially suitable for solving complex problems with internal mechanism. In this paper, BP neural network modeling method is used to model the denitrification process of SCR, and then according to the established model, neural network adaptive control is used to realize the control of nitrogen oxide concentration at the outlet of SCR. In this paper, the background and significance of the research are introduced, and the structure, algorithm and modeling steps of neural network are also introduced. Then, the mechanism of SCR denitrification process is analyzed, and the influencing factors of flue gas concentration at outlet are analyzed. The input and output variables are determined. Based on the field data of the SCR denitrification system equipped with the 600MW unit in a coal-fired power plant, the data are selected and preprocessed first, and then the appropriate neural network model structure and parameters are selected. Train the model. Then some field data are collected to verify the accuracy of the model. The results show that the error is within the acceptable range. Then the BP neural network model based on SCR denitrification process is established. The BP neural network model can accurately reflect the dynamic characteristics of denitrification system. Finally, the BP neural network model is taken as the controlled object, and the neural network model reference adaptive control method is adopted to control the NOx concentration at the outlet of SCR denitrification process. BP neural network is used as the controller to reduce the difference between the output of the controlled object and the output of the reference model by constantly modifying the weights of the neural network, so that the output of the controlled object can respond to the output of the reference model quickly. Ultimately, the control meets the desired requirements. The simulation results show that the outlet NOx concentration has good response characteristics, which shows that this control scheme is feasible and lays a foundation for its practical application.
【学位授予单位】：华北电力大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP183;TM621.8

【参考文献】