Fig. 1. Schematic diagram of FTIR measurement setup for hyperspectral emission of flames

Fig. 2. Calibration of blackbody radiation signal

Fig. 3. Calibrated flame emission spectra from experimental measurements. (a) Spectrally averaged radiative intensity; (b) distribution of spectral radiative intensity along centerline at H_AB=10 mm and H_AB=40 mm

Fig. 4. Diagram of emission spectrum measurements for axisymmetric flames

Fig. 5. Physics-based neural network for retrieving multi-parameter scalar fields in flames based on hyperspectral emission measurements

Fig. 6. Simulation results of C₂H₄-air diffusion flame. (a) 2D temperature field; (b) distributions of temperature, three gas mole fractions, and SVF at H_AB=50 mm; (c) distributions of temperature, three gas mole fractions, and SVF at H_AB=20 mm

Fig. 7. Retrieval results at H_AB=50 mm with 10% random noise in spectra. (a) Temperature; (b) mole fraction of CO₂; (c) mole fraction of CO; (d) mole fraction of H₂O; (e) SVF

Fig. 8. Retrieval results at H_AB=20 mm with 10% random noise in spectra. (a) Temperature; (b) mole fraction of CO₂; (c) mole fraction of CO; (d) mole fraction of H₂O; (e) SVF

Fig. 9. Reconstructed 2D scalar fields of experimental flame by PBNN and FTIR measurement compared with results from Liu et al.^[24], Sun et al.^[25], and U-Net^[18]. (a) Temperature; (b) CO₂ mole fraction

Fig. 10. Retrievd radial distributions of experimental flame at z=20, 30, 40 mm, and distributions of flame center line of r=0 mm by PBNN and FTIR measurement, compared with results from Liu et al.^[24], Sun et al.^[25], and U-Net^[18]. (a) Temperature; (b) CO₂ mole fraction

Fig. 11. Reconstructed 2D scalar fields of experimental flame by PBNN and FTIR measurements. (a) CO mole fraction; (b) H₂O mole fraction; (c) SVF compared with results from Sun et al.^[25]