摘要
充分利用火电厂生产运行的大量历史数据,提出了一种基于特征频段波动相似性的复杂热力系统相关信号选取方法.它针对信号变化时特征分量大多集中于某一频段的特性,利用带通滤波器消除趋势项及部分噪声干扰,通过计算归一化的滑动窗方差突出了信号特征频段内的有效信息,以便于直观判断信号间的相关性.通过对某600 MW机组数据进行实例分析,表明该方法能简捷地寻找到燃烧扰动相关信号,并能区分这些信号的相关性强弱.
Making fully use of the large amount of operation historical data from power plants, a complex thermal system correlation signals extracting method was proposed based on similarity of data fluctuation component within characteristic frequency band. Since characteristic components of correlation signals are always concentrated in a specified frequency band, a band-pass filter was designed to eliminate the trend terms and part noise disturbance. The normalization variances with moving window were calculated to en- large availability information in signals characteristic frequency band, so as to estimate the correlation of signals easily. An example analysis based on a 600 MW unit operation data shows that this method can find correlated signals of combustion disturbance effectively, and can distinguish their relative strength.
出处
《动力工程》
CSCD
北大核心
2009年第11期1004-1007,1012,共5页
Power Engineering
基金
国家自然科学基金资助项目(50776030)
关键词
火电厂
复杂热力系统
数据挖掘
相关信号
波动相似性
thermal power plant
complex thermal system
data mining
correlation signal
fluctuation similarity
