基于大数据的总排口NO_(x)浓度分析及预测  被引量:6

Big data analysis and prediction of NO_(x) concentration at the chimney

在线阅读下载全文

作  者:张军[1] 陈鸥[1] 罗志刚 杨建辉[1] 刘国栋 ZHANG Jun;CHEN Ou;LUO Zhigang;YANG Jianhui;LIU Guodong(Beijing Guodian Longyuan Environmental Engineering Co.,Ltd.,Beijing 100039,China)

机构地区:[1]北京国电龙源环保工程有限公司,北京100039

出  处:《洁净煤技术》2020年第S01期200-205,共6页Clean Coal Technology

基  金:国家能源投资集团科技创新项目(GJNY-20-29-1)

摘  要:截至2018年底,除部分地区外国内火电机组已基本完成超低排放改造。超低排放对火电机组喷氨总量控制提出了新要求,传统的控制方式由于控制精度不高、延迟较大等问题导致总排口NO_(x)难以实现稳定达标排放,或为保持总排口NO_(x)达标而过量喷氨,导致催化剂和空预器堵塞,影响装置的寿命和使用,提高了系统的运行成本。通过大数据分析方法,采用线性回归、浅层神经网络、决策树回归、极端随机树回归等4种模型,分析锅炉负荷、炉膛总风量、给煤量、一次风、二次风、燃尽风等对总排口NO_(x)的影响。计算结果表明,极端随机树回归在训练集、验证集、交叉验证集上均表现出较好的预测效果。在分析的各测点中,炉膛总风量、氨气流量、入口烟气压力、入口烟气温度、入口NO_(x)浓度、锅炉负荷、各燃烧器一次风速、各磨煤机给煤量等对总排口NO_(x)浓度的影响较大;超低排放控制需对总排口NO_(x)浓度进行精准控制时,需重点考虑这些测点。入口O_(2)浓度、省煤器出口含氧量、锅炉左右侧二次风量、前后墙燃尽风量等测点对总排口NO_(x)浓度的影响较小,在建立预测模型时可忽略。选取影响最重要70余个测点,对总排口NO_(x)进行预测,并与实际测量数据进行对比,分析各算法的预测精度及特点。对比结果表明,极端随机树分析模型预测的总排口NO_(x)值与实际测量值吻合度较高,可得到较好的预测效果。By the end of 2018,in addition to some regions,domestic thermal power units have basically completed the transformation of ultra-low emission.The ultra-low emission puts forward new requirements for the total amount control of ammonia injection in thermal power units.Due to the problems such as low control accuracy and large delay in the traditional control mode,it is difficult to achieve stable emission of NO_(x)at the chimney,or excessive ammonia injection is made to keep the NO_(x)below the standard at the chimney,resulting in blockage of catalyst and air preheater,affecting the life and use of the device,and increasing the operation cost of the system.Through the big data analysis method,linear regression,shallow neural network,decision tree regression and extreme random tree regression were used in this paper to analyze the influence of boiler load,furnace total air volume,coal supply,primary air,secondary air,burn out air and other measuring points on the chimney NO_(x).The results show that extreme random tree regression has good prediction effect on training set,verification set and cross validation set.Among the measuring points used for analysis,the total air volume,ammonia flow rate,inlet flue gas pressure,inlet flue gas temperature,inlet NO_(x)concentration,boiler load,primary air velocity of each burner and coal feeding amount of each pulverizer have great influence on NO_(x)at the general discharge outlet.When it is necessary to accurately control the total discharge NO_(x)under ultra-low emission control,these measuring points should be considered.However,the influence of O_(2)concentration at inlet,oxygen content at economizer outlet,secondary air volume at left and right sides of boiler,and burnout air volume of front and rear walls have little influence on NO_(x)at total discharge outlet,which can be ignored when establishing prediction model.On this basis,more than 70 points with the most important influence are selected to predict the chimney NO_(x).Compared with the actual measurement data,th

关 键 词:大数据 脱硝 总排口NO_(x)浓度 精细化控制 

分 类 号:X701[环境科学与工程—环境工程]

 

参考文献:

正在载入数据...

 

二级参考文献:

正在载入数据...

 

耦合文献:

正在载入数据...

 

引证文献:

正在载入数据...

 

二级引证文献:

正在载入数据...

 

同被引文献:

正在载入数据...

 

相关期刊文献:

正在载入数据...

相关的主题
相关的作者对象
相关的机构对象