Combustion disturbance is one of the most important cause leading to operation parameters' fluctuation by big margins.For signals which are easily affected by combustion disturbances

the analysis efficiency is lower only depend on mechanism analysis.Based on data-base of plant-level supervisory information system(SIS)

a data-analysis method has been put forward to extract signals correlated to combustion disturbances from a great amount of signals.The combustion distrubance components of data have been firstly obtained from band pass filtering behavior of wavelet transform

and their variances with moving windows are calculated

thereby

the correlalion of signals with combustion disturbance can be measured quantitatively according to the correlation coefficient between said variance vectors.Verification carried out on one 600 MW unti shows that the said method is comparatively effective

being simple in computation

and easy in realization.