摘要
燃煤机组NO_x生成量的准确测量是改善脱硝控制品质的关键因素。针对电站锅炉长期处于变工况的动态过程,并考虑到辅助变量存在的时间异步以及NO_x生成量测量存在的迟延问题,提出基于最小二乘支持向量机与互信息(MI-LSSVM)的燃煤机组NO_x生成量的动态软测量方法。首先,介绍了数据预处理及辅助变量的选择;其次,根据理论知识及实际情况确定NO_x生成量的测量迟延,以互信息最大化为准则,确定各个辅助变量的时间迟延;最后,将包含迟延信息的辅助变量数据与过去时刻的NO_x测量数据作为输入,建立最小二乘支持向量机(LSSVM)模型,实现动态软测量。基于某电厂330 MW机组的一段历史运行数据,验证了模型的预测效果。验证结果表明:基于MI-LSSVM的模型精度较单独使用LSSVM模型有明显的提高,且MI-LSSVM模型具有一定的提前预测功能。该模型不仅具有良好的预测效果,而且为脱硝系统喷氨量的准确控制奠定了良好的基础。
The accurate measurement for amount of NO x production of coal-fired units is the key factor to improve denitration control system.Aiming at the dynamic process of power plant boilers in long-term changing conditions and considering the time asynchronous of auxiliary variables and the delay problem of NO x production measurement,the dynamic soft-sensing method for NOx production amount of coal-fired units based on least squares support vector machine and mutual information(MI-LSSVM)is proposed.Firstly,the selection of data preprocessing and auxiliary variables is introduced.Secondly,based on theoretical knowledge and actual conditions,the measurement delay of NOx production is selected;based on the criterion of maximizing the mutual information,the time delay of each auxiliary variable is determined.Finally,the auxiliary variable data including delay information and the measurement data of NOx in the past are taken as input,and the least squares support vector machine(LSSVM)model is established to realize dynamic soft-sensing.Based on some of the historical operation data of a certain330MW generation unit,the prediction effect of the model is verified.The verified result shows that the accuracy of the model based on MI-LSSVM is significantly better than that of using LSSVM model,and the MI-LSSVM model has a certain predictive function.The method has good prediction effect and establishes a basis for accurate control of ammonia injection in the denitration system.
作者
赵征
袁洪
宋选锋
ZHAO Zheng;YUAN Hong;SONG Xuanfeng(School of Control and Computer Engineering,North China Electric Power University,Baoding 071003,China)
出处
《自动化仪表》
CAS
2018年第12期31-34,38,共5页
Process Automation Instrumentation
基金
新能源电力系统国家重点实验室开放课题基金资助项目(LAPS16008)
中央高校基本科研业务费专项资金基金资助项目(2017MS133)
关键词
燃煤机组
NOx
生成量
动态软测量
主元分析
最小二乘支持向量机
互信息
Coal- fired units
Amount of NO x production
Dynamic- soft sensing
Principal component analysis
Least squares support vector machines(LSSVM)
Mutual information(MI)
作者简介
赵征(1978-),女,博士,副教授,主要从事大型机组智能优化控制方向的研究,E-mail:zheng_zhao@ncepu.edu.cn;袁洪(通信作者),女,在读硕士研究生,主要从事热工过程控制及建模方向的研究,E-mail:yuanhong_ncepu@126.com.
