论文部分内容阅读
针对传统的热工信号相关性分析无法兼顾整体趋势相关性和局部波动相似性分析的不足,利用小波变换的多分辨率分析思想,提出从不同频率尺度来研究信号的相关性。通过小波变换的多层分解与重构得到不同频率范围内的信号分量,依据同一频率信号的波动相似性,计算相关系数来定量描述该尺度下信号间的关联程度。对某600MW机组数据进行实例分析,表明该方法不仅可以定量分析信号低频趋势的相关性强弱,同时也能够挖掘出高频波动相似性强的信号,从而拓展了热工相关信号的挖掘范围。
In view of the traditional correlation analysis of thermal signals can not take into account the overall trend correlation and local fluctuation similarity analysis, using the multi-resolution analysis of wavelet transform, it is proposed to study the signal correlation from different frequency scales. The signal components in different frequency ranges are obtained by multi-layer decomposition and reconstruction of wavelet transform, and the correlation coefficients are calculated according to the fluctuation similarity of the same frequency signal to quantitatively describe the correlation between the signals under this scale. An example analysis of a 600 MW unit data shows that this method can not only quantitatively analyze the correlation of the low-frequency trend of signals, but also excavate signals with similar high-frequency fluctuations, thereby expanding the scope of the thermal-related signal mining.