The used instrumentation for combustion diagnostics can be divided into low-frequency (in the range of 1 Hz) and high-frequency (in the range of 10 kHz) data acquisition systems. The former group involves sensors which are typical for characterizing the overall operation or describe the steady-state conditions due to the large integration time. These are starting from the simple thermocouples and resistance thermometers and continue with pressure and flow meters. The flue gas analyzer and the flame emission spectrometer also fall into this category.
There are two high-frequency device types available in our lab for capturing the transient operation of the flame. They are microphones and photomultiplier tubes. Data acquisition is performed phase-locked at maximum frequency of 105 kHz.
The low-frequency signal is always evaluated in the temporal space, and their result is straightforward to evaluate. Nevertheless, high-frequency temporal signals capture rich details of the ongoing phenomena which enable the use of various spectral and statistical tools to gather indirect information as well.
For spectral analysis, Fast Fourier transform and Short-time Fourier transform are classical tools which are quick algorithms, and the results are easy to read in a general case. However, combustion noise is tricky since the low-frequency peaks are mostly well-localized while the temporal evolution of the broadband roar also contains relevant information. Therefore, another technique can be perfectly used for this purpose which is the wavelet transform.
The evaluation of both the temporal and spectral data can be done with statistical techniques. Starting from the average and histograms, joint probabilities and variance analysis also provide detailed information for diagnostic purposes.