综合智慧能源 ›› 2022, Vol. 44 ›› Issue (3): 63-69.doi: 10.3969/j.issn.2097-0706.2022.03.010

• 智能电力 • 上一篇    下一篇

基于数据驱动的CFB机组变负荷工况SO2质量浓度建模

李彩霞1(), 赵军1,*(), 李建伟1(), 王伟1(), 王杰2(), 于浩洋3()   

  1. 1.内蒙古蒙泰不连沟煤业有限责任公司煤矸石热电厂,内蒙古 准格尔 010321
    2.内蒙古蒙泰不连沟煤业有限责任公司,内蒙古 准格尔 010321
    3.华北电力大学 控制与计算机工程学院,北京 102206
  • 收稿日期:2021-09-25 修回日期:2022-01-22 出版日期:2022-03-25 发布日期:2022-03-28
  • 通讯作者: 赵军
  • 作者简介:李彩霞(1981),女,工程师,从事火电厂热工检修工作, yuhaoyang11111@126.com
    李建伟(1985),男,工程师,从事电厂运行管理工作, 44137061@qq.com
    王伟(1993),男,助理工程师,从事电厂技术监督管理工作, 1135781143@qq.com
    王杰(1972),男,助理工程师,从事电力生产和管理工作, 277873238@qq.com
    于浩洋(1996),男,在读博士研究生,研究方向为循环流化床机组污染物排放控制, 120202127004@ncepu.edu.cn
  • 基金资助:
    中国华电集团科技项目(CHDKJ21-02-161)

Data-driven modeling for SO2 mass concentration of CFB units under variable load conditions

LI Caixia1(), ZHAO Jun1,*(), LI Jianwei1(), WANG Wei1(), WANG Jie2(), YU Haoyang3()   

  1. 1. Gangue Thermal Power Plant of Inner Mongolia Mengtai Buliangou Coal Industry Company Limited,Jungar 010321,China
    2. Inner Mongolia Mengtai Buliangou Coal Industry Company Limited,Jungar 010321,China
    3. School of Control and Computer Engineering,North China Electric Power University,Beijing 102206,China
  • Received:2021-09-25 Revised:2022-01-22 Online:2022-03-25 Published:2022-03-28
  • Contact: ZHAO Jun

摘要:

由于循环流化床(CFB)机组在动态过程中缺乏有效的污染物生成与还原模型的指导,导致变负荷能力在一定程度上受到污染物排放水平的制约。提出了一种基于数据驱动的循环流化床机组SO2质量浓度动态模型,应用极限学习机建立基础模型,根据循环流化床污染物生成与还原机理,选择合适的输入变量,并应用遗传算法对该模型加以改进,使该模型具有较高的精度,并在动态工况下有较好的建模结果。该模型可以为SO2质量浓度控制系统提供有效指导。同时,在所提出的模型基础之上,在智能平行控制理论框架下,虚拟系统与实际系统相结合形成平行系统,提出了循环流化床机组SO2控制系统智能平行控制方法,可为今后循环流化床机组SO2低排放智能控制提供参考,在一定程度上有利于提升循环流化床机组变负荷能力。

关键词: 循环流化床, 极限学习机, 遗传算法, SO2质量浓度动态模型, 智能平行控制, 数据驱动, 智能发电

Abstract:

Due to the lack of effective guidance for pollutant generation and reduction modelling in the dynamic process of circulating fluidized bed(CFB)units,the load adjusting capacity is restricted by pollutant emission to a certain extent.A data-driven SO2 concentration dynamic model for CFB units is established based on an extreme learning machine.According to the generation and reduction mechanisms of pollutants from CFBs,appropriate input variables are selected.The model improved by genetic algorithms is of higher accuracy and better modelling results under dynamic conditions.The model can provide effective guidance for SO2 concentration control systems.At the same time,a parallel system integrating a real system with its virtual counterpart is made on the basis of the proposed model under the framework of intelligent parallel control theory.The proposed intelligent parallel control for the SO2 control systems can provide references for the SO2 emission control of the following CFBs for boosting their load adjusting capacities to a certain extent.

Key words: CFB, extreme learning machine, genetic algorithm, SO2 concentration dynamic model, intelligent parallel control, data driven, intelligent power generation

中图分类号: