热工过程数据聚类分析研究及其在模型辨识中的应用  

STUDY ON CLUSTER ANALYSIS OF THERMODYNAMIC DATA AND ITS APPLICATION IN MODEL IDENTIFICATION

作　　者：李涛永[1] 刘长良[2] 闫华光[1] 

机构地区：[1]中国电力科学研究院,北京100192 [2]华北电力大学控制与计算机工程学院,河北保定071003

出　　处：《热力发电》2012年第6期24-27,共4页Thermal Power Generation

基　　金：国家863资助项目(SQ2010AA1120919001);中国电力科学研究院青年基金项目(JL84-11-002)

摘　　要：发电厂热工过程具有非线性、大惯性、大延迟的特性,利用欧式谱系聚类分析方法对热工过程进行工况划分,将热工过程的非线性转化为若干个工况点的线性化问题,并把时间戳属性引入聚类分析的数据矩阵中。利用改进微分进化算法对给煤量设定值、主蒸汽压力的辨识结果表明,聚类分析方法能够提高模型辨识中工况划分的合理性,并且在给煤量设定值阶跃扰动时使负荷相对稳定。根据试验数据辨识得到的模型参数存在差异。The thermodynamic process in thermal power plants boasts features of non - liearity, large in- ertia,and long time delay. By using cluster analysis method in Euro - mode spectrum system, the per- formance of thermodynamic process has been divided, and the nonlinearity of thermodynamic process having been turned into a problem of linearization at some performance points,and the time stamps be- ing introduced into the data matrix of cluster analysis. The results in identification of set value for coal feed and main steam pressure while using modified differential evolution method show that the cluster analysis can enhance the rationality of performance division in model identification, and making the load to be relatively stable under stepped disturbance of the coal feed set value, but difference still exists in model parameters obtained from identification according to the test data.

关 键 词：热工过程 聚类分析 给煤量 主蒸汽压力 模型辨识 

分 类 号：TM621[电气工程—电力系统及自动化]