Diagnosis of Combustion Instability Using Waveform Pattern of Dynamic Pressure Signal

This study introduces a new concept that can evaluate the combustion status from the waveform patterns of the dynamic pressure signal, and an initial investigation result on the possibility of combustion instability monitoring, by classifying the waveform of time-series discrete dynamic pressure data, measured in a gas turbine combustor. Based on variations in amplitude among four continuously adjacent data points of dynamic pressure, there are 24 waveform patterns that can be classified. These patterns can be further streamlined to derive three representative diagnostic factors. The representative factors tended to converge to the specific values when combustion instability occurred. Furthermore, distinct characteristic value distributions were observed in stable and transitional states, respectively. These results implied that the waveform pattern, newly defined in this study, can be effectively used as a combustion instability diagnostic tool.