Spectral analysis of combustion noise and flame pattern recognition

In many situations, flame patterns in a combustion chamber cannot be observed directly by using an electronic or optical probe. However, an experienced engineer can identify the burning process by listening to the noise that it generates. In this paper, we study acoustic characteristics of turbulent impinging flames by using spectral analysis and statistical pattern recognition. By experimenting with the ignition method, different flame patterns were generated in a laboratory. We find that each flame pattern can be characterized effectively by using the power spectrum of the noise and can be identified by using this information alone.