引用本文:牛玉广,潘岩,陈曦.选择性催化还原烟气脱硝深度结构和深度控制[J].控制理论与应用,2019,36(1):65~72.[点击复制]
NIU Yu-guang,PAN Yan,CHEN Xi.Depth structure control and depth control of selective catalytic reduction flue gas denitration system[J].Control Theory and Technology,2019,36(1):65~72.[点击复制]
选择性催化还原烟气脱硝深度结构和深度控制
Depth structure control and depth control of selective catalytic reduction flue gas denitration system
摘要点击 2722  全文点击 1230  投稿时间：2018-01-15  修订日期：2018-09-13
查看全文  查看/发表评论  下载PDF阅读器
DOI编号  10.7641/CTA.2018.80043
  2019,36(1):65-72
中文关键词  SCR烟气脱硝  深度控制  模拟退火  单神经元自适应  粒子群  优化
英文关键词  selective catalytic reduction(SCR) flue gas denitration  depth control  simulated annealing  single neuron adaptive  particle swarm optimization  optimization
基金项目  
作者单位E-mail
牛玉广 华北电力大学控制与计算机工程学院 nyg_ncepu@sina.com 
潘岩* 华北电力大学控制与计算机工程学院 1531920937@qq.com 
陈曦 华北电力大学控制与计算机工程学院  
中文摘要
      针对火力发电厂氮氧化物排放环保要求日渐严格的现状, 以优化排污控制效果为目标, 提出了于喷氨格栅加装 调节阀和测点的优化改造方案. 在此基础上参考受限玻尔兹曼机网络结构设计了多输出串级平面场深度控制结构方案; 为实现选择性催化还原(selective catalytic reduction, SCR)脱硝系统的深层优化, 提出单神经元自适应–模拟退火(single neuron adaptive simulated annealing, SNASA)算法, 并作为优化方案的控制器对喷氨系统进行深度结构控制; 提出深度 粒子群(deep particle swarm optimization, DPSO)算法, 在相同方案中作为控制器以实现喷氨系统的深度控制. 设计了控 制品质系数, 为各系统的比较提供直观依据. 仿真结果表明: 所设计的两个基于深度网络的喷氨优化系统控制品质良好, 鲁棒性强, 较基于传统方案的喷氨控制优势明显, 对工程现场有一定的指导意义
英文摘要
      In view of the situation that limitation of nitrogen oxide created by thermal power plant becomes tighter than ever, an optimum proposal of adding control valves and concentration transmitters are presented to improve the pollution control quality. Refer to the network structure of restricted Boltzmann machine, a multi-output cascade field control strategy based on the hardware design is given. It is proposed that the single neuron adaptive simulated annealing (SNASA) algorithm is used as optimum controllers of the flue gas denitration control system to achieve better control quality by exertion deep structure. It is presented that the deep particle swarm optimization (DPSO) algorithm is used as the same controllers to have depth control. Control quality factor is designed in order to compare between different systems. The simulation results indicate that the two presented control systems based on deep networks have better control qualities and stronger robustness than traditional control system, and have the practical significance.