发电技术 ›› 2019, Vol. 40 ›› Issue (3): 246-252.DOI: 10.12096/j.2096-4528.pgt.19102

• 燃煤发电系统能源高效清洁利用 • 上一篇    下一篇

基于遗传算法优化BP神经网络的SCR脱硝系统催化剂体积设计

唐诗洁1(),陆强1(),曲艳超2,任翠涛2,杨勇平1   

  1. 1 生物质发电成套设备国家工程实验室(华北电力大学), 北京市 昌平区 102206
    2 北京华电光大环境股份有限公司, 北京市 昌平区 102206
  • 收稿日期:2019-01-25 出版日期:2019-06-30 发布日期:2019-07-02
  • 作者简介:唐诗洁(1993),女,硕士研究生,研究方向为火电厂脱硝综合治理, tangsj1120@126.com|陆强(1982),男,教授,博士生导师,主要从事固体燃料高效热利用以及烟气污染物治理方面研究工作, qianglu@mail.ustc
  • 基金资助:
    国家重点基础研究发展计划项目(2015CB251501);北京市科技新星(Z171100001117064);中央高校基本科研业务费专项资金(2018ZD08);中央高校基本科研业务费专项资金(2016YQ05)

Catalyst Volume Design in SCR Denitrification System Based on Genetic Algorithm Optimized BP Neural Network

Shijie TANG1(),Qiang LU1(),Yanchao QU2,Cuitao REN2,Yongping YANG1   

  1. 1 National Engineering Laboratory for Biomass Power Generation Equipment, North China Electric Power University, Changping District, Beijing 102206, China
    2 Beijing National Power Group Co., Ltd., Changping District, Beijing 102206, China
  • Received:2019-01-25 Published:2019-06-30 Online:2019-07-02
  • Supported by:
    National Basic Research Program of China(2015CB251501);Beijing Nova Program(Z171100001117064);Fundamental Research Funds for the Central Universities(2018ZD08);Fundamental Research Funds for the Central Universities(2016YQ05)

摘要:

火电厂SCR脱硝系统的设计需要在满足脱硝效率的同时,尽可能节约成本,因此需要准确预测SCR脱硝所需的催化剂体积。火电厂的烟气条件复杂多变,烟气温度、烟气流量、出入口NOx浓度等参数都会影响SCR催化剂的体积设计,因此催化剂体积预测是一个多因素耦合的问题。针对这一特点,使用BP神经网络对催化剂体积设计进行了预测,并针对该模型结构上的缺陷,进行基于遗传算法优化的神经网络建模研究。结果表明,遗传算法优化后的BP神经网络模型预测精度和数据拟合能力均有提高,为脱硝系统的催化剂体积设计提供了新思路。

关键词: SCR催化剂, 催化剂体积预测, BP神经网络, 遗传算法

Abstract:

The design of the SCR denitrification system in coal-fired power plants requires the efficient denitrifi-cation efficiency and the low cost. Hence, it is essential to accurately calculate the volume of SCR denitrification catalysts. The flue gas conditions of thermal power plants are complex and changeable. Flue gas temperature, flue gas flow, inlet and outlet NOx concentrations, and other parameters all affect the volume of the SCR catalyst.Therefore, catalyst volume prediction is a multifactor coupling problem. For this feature, the BP neural network model was used to predict the volume design of the catalyst, and the neural network modeling based on genetic algorithm optimization was investigated for the structural defects of the BP neural network model. The results show that the prediction accuracy of BP neural network model optimized by genetic algorithm is promising, which provides a new way for catalyst volume design of SCR denitrification.

Key words: SCR catalyst, catalyst volume prediction, BP neural network, genetic algorithm