基于现场数据和主元分析的主汽温系统多变量建模  

Multivariable modeling of the main steam temperature based on field data and PCA

作　　者：侯晓宁 

机构地区：[1]华电郑州机械设计研究院有限公司,郑州450046

出　　处：《华电技术》2018年第5期25-29,32,共6页HUADIAN TECHNOLOGY

摘　　要：针对主汽温系统具有的多变量、强耦合的特点,采集大量现场运行数据,利用主元分析原理,从影响主汽温的11个过程变量中提取5个主要变量,作为模型输入,同时,将机理建模、试验建模和智能优化算法建模相结合,利用现场数据建立主汽温系统多入单出的多变量模型,解决了在生产现场不允许或没有条件进行大范围频繁阶跃扰动试验的问题,克服了多变量智能寻优时,初始值范围和模型结构不确定的困难,建立出可以用于主汽温系统控制研究的多变量模型。As the main steam temperature system is a multivari-able and strong coupling system,a large number of field operational data is collected. Through principal component analysis(PCA),five main variables are extracted from 11 procedure variables as the input of the main steam temperature system. Simultaneously,the mechanism modeling,experimental modeling and intelligent optimization algorithms modeling are combined. Finally,the multiple-input single-output multivariable model is established by using the field data. This solution solves the problem of frequent and large-scale step disturbance tests in which it is not allowed or cannot to get to the productive site. It overcomes the uncertainty of the range of initial values and model structure for multivariable intelligent optimizing,and established a multivariable model that can be used for studying main steam temperature system control.

关 键 词：主汽温 多变量建模 主元分析 现场数据 

分 类 号：TM621[电气工程—电力系统及自动化]