Multi-speciation in shock tube kinetics using deep neural networks and cavity-enhanced absorption spectroscopy

Abstract

Chemical kinetic experiments of fuel oxidation/pyrolysis are quite complicated with a multitude of species being formed and consumed. It is desirable to have a diagnostic strategy that can detect many species simultaneously with high sensitivity, selectivity, and fast time response. In this work, cavity-enhanced absorption spectroscopy (CEAS) and deep neural network (DNN) are exploited for selective and simultaneous multi-species detection in high-temperature shock-tube experiments. As a representative case, time-histories of major products of propylbenzene pyrolysis are measured behind reflected shock waves at T 950–1300 K and P 1 atm. A distributed feedback inter-band cascade (ICL) laser emitting near is used as the laser source. Laser wavelength tuning over 3038.6–3039.8 cm and denoising models based on DNN are employed to differentiate the broadband absorbance spectra of benzene, toluene, ethylbenzene, ethylene, styrene and propylbenzene. The models are able to clean noisy absorbance spectra and split these into contributions from reference species by multidimensional linear regression (MLR). Off-axis CEAS is implemented to improve sensitivity to weak absorbers by amplifying effective laser path-length. To the best of our knowledge, this work reports the first successful implementation of time-resolved multispecies detection with a single narrow wavelength-tuning laser and CEAS configuration. This work also represents the first study of simultaneous measurement of multiple species during propylbenzene pyrolysis using laser absorption spectroscopy.