A Detection Method for Lean Blowout Precursors of Bluff-Body Premixed Flames Based on the Data Balance Ratio of Electrostatic Signal

Lean combustion, while fuel-efficient, exhibits inherent instability with a high propensity for blowout. To mitigate this issue, bluff-body flame holders are conventionally employed in afterburners and gas turbine combustion chambers to stabilize lean flames. Nevertheless, these devices do not entirely eliminate the risk of blowout. Early detection of blowout precursors is critical for enabling timely warnings and ensuring flame combustion stability. This paper introduces a novel lean blowout precursory detection method that synergizes data balance ratio analysis with non-contact electrostatic sensing technology. The electrostatic sensor is used to capture the movement of charged particles generated by flame ionization to obtain the electrostatic signal. The reflection of three common lean blowout precursory indexes on flame combustion state is analyzed. The three common lean blowout precursory indexes include the ratio of standard deviation to mean value, low-frequency power ratio and skewness coefficient. It is found that only the skewness coefficient of the above three indexes can basically distinguish the near lean blowout state and stable combustion state. However, it is possible that the stable combustion state may be misjudged as the near lean blowout state. The probability distribution of the electrostatic signal amplitude is analyzed. The cause of the misjudgment is the asymmetric probability distribution of data points relative to the mean value of the electrostatic signal in the stable combustion state. A lean blowout precursory detection method based on data balance ratio of electrostatic signal is proposed. Compared to the three common lean blowout precursory indexes, the accuracy of the data balance ratio index is 0.941, specificity is 0.933, and the AUC value is 0.973, all of which are the highest values. Meanwhile, its false-alarm rate is only 0.059, achieving the lowest level. Using the Rayleigh distribution as the boundary for combustion state classification, the threshold value of 0.101 for the data balance ratio index is established through theoretical calculations, enabling effective discrimination between the stable combustion state and the near lean blowout state.