A multi-scale analysis approach to selecting signals related with combustion status

In order to improve the diagnostic accuracy of boiler combustion status, it is necessary to select more signals related with combustion status than flame signal and furnace pressure signal. However, it is inefficient only using mechanism analysis when selecting because of the complicated input-output mapping relation of boiler combustion. Based on multi-scale analysis, a new method selecting related signal is presented according to the fluctuation similarity of thermal signals. Firstly, components of different frequency ranges are obtained via multi-level wavelet decomposition and reconstruction. Then, correlation coefficients are calculated based on the assumption of fluctuation similarity between components in same frequency range, which can be used to describe the correlation degree of those components. By this method, the signals related with combustion status can be selected through calculating the correlation coefficients of their high frequency components. The calculation results using on-site data prove the validity of the proposed method.