基于火焰结构特征提取的振荡燃烧故障快速检测技术

With the development of advanced gas turbines towards high load and low emission， the probability of inducing combustion instability in the combustion chamber continues to increase. It has been long desired to develop new detection technologies for accurate and rapid detection of combustion instability faults. In this paper， a rapid detection technique of combustion instability based on flame chemiluminescence image acquisition， proper orthogonal decomposition feature extraction and continuous wavelet transform time-frequency analysis is proposed. The technique can resolve the frequency domain and its time evolution simultaneously. Through experimental verification， it is found that for oscillations to stability， stability to oscillation， one main frequency to another main frequency and other mode transition processes， the mode time coefficients obtained by using the proper orthogonal decomposition of flame images can obtain faster detection speed than the pressure signal. The improvement of the detection speed is more than 60% in the tested cases.