华电技术 ›› 2021, Vol. 43 ›› Issue (5): 9-14.doi: 10.3969/j.issn.1674-1951.2021.05.002

• 智能电力 • 上一篇    下一篇

基于IQGA-GRNN模型的SCR脱硝出口NOx质量浓度预测方法

曹喜果1(), 张永涛1,*(), 李雅恬2()   

  1. 1.新疆工程学院 能源工程学院,乌鲁木齐 830091
    2.河北华电石家庄鹿华热电有限公司,石家庄 050000
  • 收稿日期:2021-01-21 修回日期:2021-04-08 出版日期:2021-05-25 发布日期:2021-05-18
  • 通讯作者: 张永涛
  • 作者简介:曹喜果(1989—),女,河南平顶山人,讲师,硕士,从事电站机组建模及仿真等方面的工作(E-mail:398606950@qq.com)。
    李雅恬(1991—),女,河北石家庄人,工程师,硕士,从事火电厂脱硝系统建模及优化等方面的工作(E-mail:365605050@qq.com)。
  • 基金资助:
    新疆维吾尔自治区高校科研计划自然科学青年研究项目(XJEDU2018Y054)

Prediction method for NOx discharged from SCR denitrification systems based on IQGA-GRNN model

CAO Xiguo1(), ZHANG Yongtao1,*(), LI Yatian2()   

  1. 1. College of Energy Engineering, Xinjiang Institute of Engineering, Urumqi 830091,China
    2. Hebei Huadian Shijiazhuang Luhua Thermal Power Company limited, Shijiazhuang 050000,China
  • Received:2021-01-21 Revised:2021-04-08 Online:2021-05-25 Published:2021-05-18
  • Contact: ZHANG Yongtao

摘要:

针对选择性催化还原(SCR) 烟气脱硝系统工艺复杂、非线性等特点,提出了一种基于改进量子遗传算法(IQGA)和广义回归神经网络(GRNN)的燃煤电站NOx排放数学模型:先采用动态旋转门对量子遗传算法(QGA)进行改进,使其搜索更为精细,然后应用IQGA对GRNN中的光滑因子进行寻优,使该算法逼近能力更强。以某300 MW供热机组为例,针对现场实际运行数据,采用IQGA-GRNN进行训练建模,并将该模型与GRNN模型、QGA-GRNN模型的预测结果进行对比,结果表明,IQGA-GRNN模型的预测值与实测值最大误差在8.0%以内,平均误差在0.2%以内,可为后续喷氨量的精准控制提供有力的支撑。

关键词: 改进量子遗传算法, 广义回归神经网络, 燃煤电站, 脱硝系统, NOx质量浓度, 预测

Abstract:

According to the complex craftwork and non-linear characteristic of SCR denitrification systems, a mathematical model for NOx discharged from coal-fired power plants was created based on Improved Quantum Genetic Algorithm (IQGA) and General Regression Neural Network (GRNN). Firstly, QGA was modified by revolving door to get the search results more accurate. Secondly, the smoothness factor in GRNN was optimized to improve the approximation ability of the algorithm. Taking a 300 MW heat-supply unit as an example, an IQGA-GRNN model was created and trained by the training data of the unit. The maximum error between the predicted value made by the model and the measured value is within 8.0%, and the average error is within 0.2%. The IQGA-GRNN model is supportive for precise control on NH3 spray.

Key words: IQGA, GRNN, coal-fired power plant, denitrification system, NOx mass concentration, prediction

中图分类号: