Flame equivalence ratio measurement using data fusion based on laser-induced plasma spectra and acoustic signals

Abstract

Laser-induced breakdown spectroscopy (LIBS) has been proven to be employed in combustion diagnostics. However, the detection performance of flame equivalence ratios by LIBS could be further improved, since the equivalence ratio is an important parameter for monitoring and optimizing combustion processes. In this work, we propose a measurement method of the flame equivalence ratio based on data fusion by combining laser-induced plasma spectra and acoustic signals. An experimental platform for simultaneous detection of plasma spectra and acoustic signals was established to study the spectral and acoustic characteristics of plasmas in flames with different equivalence ratios. The characteristic lines of C I 247.86 nm, CN (0–0) 388.29 nm, H I 656.28 nm, N I 746.83 nm and O I 777.42 nm were selected for spectral analysis. A gradual increasing and then decreasing trend of line intensities with delay time was obtained, while line intensities gradually increase with the laser energy. It is shown that the optimal spectral detection conditions are a laser energy of 85 mJ and a delay time of 200 ns for obtaining high-quality spectral signals. In addition, the dependence of the flame plasma acoustic signals on detection angles and detection distances was also investigated, with the optimal detection conditions at a detection angle of 30° and a distance of 10 cm. It is found that the spectral line intensity ratios and the peak-to-peak values of acoustic signals both have a strong linear relationship with the flame equivalence ratios, enabling them to be used for the measurement of the equivalence ratio. The prediction performances for flame equivalence ratios of five different methods, including the standard curve method with spectra, PLS-DA with spectra, standard curve method with acoustic signals, PLS-DA with acoustic signals and data fusion, were compared. It is worth noting that the utilization of the data fusion strategy obviously improves the prediction accuracy by integrating the spectral and acoustic features of the plasma, enabling the prediction error of the validation samples to be reduced by 3 times. The present results demonstrate the effectiveness of the data fusion strategy combining spectral and acoustic signals of laser-induced plasmas for the accurate measurement of flame equivalence ratios, which plays an important role in combustion diagnostics.