陡河发电厂200MW锅炉燃烧优化神经网络建模

发布时间：2019-01-19 18:17

【摘要】：随着我国日益严峻的能源和环境形势,燃煤电厂面临着越来越大的节能和减排压力。电厂锅炉作为其主要耗能设备,也受到越来越多的关注。首先是我国电厂锅炉的运行效率低于国际先进水平,相比于日本等发达国家,我国火力发电需要消耗巨大的能源。其次,在电力调峰的过程中,锅炉可能处于低负荷运行,为了保证经济效益,针对锅炉效率的研究是经济发展的必然要求。再次,火电锅炉燃烧产生大量NOx等污染物,造成巨大的环境负担。在当前条件下,锅炉燃烧控制系统技术上的不足越来越突出,难以满足日益增长的要求。在这种背景下,效率高,NOx排放低的火电锅炉燃烧优化技术成为了研究人员关注的热点。本文在深入研究现有的燃烧优化方法的基础上,分析了神经网络在建立热电锅炉燃烧模型方面具有的特点并结合陡河电厂200MW热电锅炉工作的实际情况,研究了运用遗传算法寻优热电锅炉燃烧参数的问题,主要研究内容如下：(1)对锅炉结构和燃烧工艺进行了详细的分析,在原有经验方法的基础上,深入理解燃烧工艺过程和电厂锅炉控制关键因素及其对锅炉燃烧性能影响方式。并从设备层面上确定了研究内容,为后续的理论分析奠定了基础。(2)利用神经网络建模和遗传算法进行模型求解,并得到优化结果。介绍了神经网络模型工作的基本原理及其工作方式,在着重介绍BP神经网络、径向基神经网络及其算法流程的基础上,利用BP网络和径向基神经网络对锅炉燃烧系统建立模型并对仿真结果进行了分析和比较,并最终确定了基于径向基网络的建模方法描述锅炉燃烧模型。利用交叉试验确定了模型的输入和输出参数。并利用遗传算法对锅炉的燃烧模型进行求解。在模型建立与寻优的章节中,总结了各个神经网络模型具有的特点及其应用场景。并利用遗传算法对燃烧系统的控制参数进行寻优,得到了提高燃烧系统热效率并降低氮氧化物排放量的参数优化值。(3)使用MFC开发了锅炉燃烧控制参数优化软件,介绍了主要界面和详细使用流程及注意要点。并将开发的计算系统应用于锅炉的实时控制。实际结果表明利用基于神经网络模型和遗传算法优化的算法在实际应用中有巨大作用,锅炉系统热效率和NOx排放量均得到优化,带来了巨大的经济效益和环境效益。最后对本文研究的不足之处做了总结并提出了展望。
[Abstract]:With the increasingly severe energy and environmental situation in China, coal-fired power plants are facing more and more pressure of energy saving and emission reduction. As the main energy consuming equipment, power plant boiler is paid more and more attention. First of all, the operating efficiency of power plant boilers in China is lower than the advanced level in the world. Compared with the developed countries such as Japan, the thermal power generation in our country needs a huge amount of energy consumption. Secondly, in the process of power peak shaving, the boiler may be in low load operation. In order to ensure economic benefit, the research on boiler efficiency is the inevitable requirement of economic development. Thirdly, the combustion of thermal power boiler produces a large amount of pollutants, such as NOx, resulting in a huge environmental burden. Under the present condition, the technical deficiency of boiler combustion control system is more and more prominent, and it is difficult to meet the increasing demand. In this context, high efficiency and low NOx emissions of thermal power boiler combustion optimization technology has become the focus of attention. On the basis of deeply studying the existing combustion optimization methods, this paper analyzes the characteristics of the neural network in establishing the combustion model of the thermoelectric boiler and combines the actual situation of the work of the 200MW thermoelectric boiler in Douhe Power Plant. The problem of optimizing combustion parameters of thermoelectric boiler by genetic algorithm is studied. The main contents are as follows: (1) the boiler structure and combustion technology are analyzed in detail. Deeply understand the combustion process and the key factors of boiler control in power plant and its influence on boiler combustion performance. The research content is determined from the equipment level, which lays a foundation for the subsequent theoretical analysis. (2) the neural network modeling and genetic algorithm are used to solve the model, and the optimization results are obtained. In this paper, the basic principle and working mode of neural network model are introduced. On the basis of introducing BP neural network, radial basis function neural network and its algorithm flow, BP neural network and radial basis function neural network are used to model the boiler combustion system, and the simulation results are analyzed and compared. Finally, the modeling method based on radial basis function network is determined to describe the boiler combustion model. The input and output parameters of the model are determined by cross test. The combustion model of boiler is solved by genetic algorithm. In the chapter of model establishment and optimization, the characteristics and application scenarios of each neural network model are summarized. The optimized parameters of combustion system are obtained by genetic algorithm. (3) the optimization software of boiler combustion control parameters is developed by using MFC, which can improve the thermal efficiency of combustion system and reduce the emission of nitrogen oxide. This paper introduces the main interface and detailed usage flow and points for attention. The developed computing system is applied to the real-time control of boiler. The practical results show that the optimization algorithm based on neural network model and genetic algorithm plays an important role in practical application, and the thermal efficiency and NOx emission of boiler system are optimized, which brings great economic and environmental benefits. At last, the deficiency of this paper is summarized and the prospect is put forward.
【学位授予单位】：华北电力大学(北京)
【学位级别】：硕士
【学位授予年份】：2016
【分类号】：TM621.2;TP183

【相似文献】