摘要
增加烟气中一氧化碳含量作为电厂锅炉燃烧调整的参数,可以有效避免局部不完全燃烧,提高锅炉经济性。因此,如何快速准确地反映烟气中一氧化碳含量至关重要。为此采用软测量方法,利用与之相关的一些参数作为输入变量,对一氧化碳含量进行BP神经网络建模。根据对模型的训练仿真和验证样本的相对误差计算,该模型具有较好地泛化能力,可以较准确地预测一氧化碳含量。利用该模型预测结果可以快速调整燃烧工况,得到CO/O2变化关系,使锅炉处于最佳过量空气系数下。
Adding the levels of carbon monoxide in flue gas as an adjustment parameter of power plant boiler combustion, can effectively avoid incomplete combustion and improve the economy of boilers. Therefore, how to quickly and accurately reflect the content of carbon monoxide in flue gas is essential. In this paper, soft measurement method is used which chooses some related parameters as input variables,establishes BP neural net- work model of carbon monoxide measurement. As the result of the training and proving process of the BP neural network of carbon monoxide measurement shows, the neural network is of good generalization, can predict the content of carbon monoxide in flue gas. It can guide operators to quickly and efficiently adjust the combustion of boilers, get the relationship between oxygen and carbon monoxide, and finally make the boilers running in optimum excessive air coefficient.
出处
《东北电力大学学报》
2014年第3期85-88,共4页
Journal of Northeast Electric Power University
关键词
软测量
BP神经网络
一氧化碳
Soft measurement method
BP neural network
Carbon monoxide
