Fig. 1. Schematic of temperature and CO₂ concentration fields measurements for axisymmetric flames

Fig. 2. Schematics of traditional measurement method. (a) Reconstruction of temperature and CO₂ concentration fields based on axial and radial laser spectral absorption measurements; (b) reconstruction of temperature and CO₂ concentration fields based on axial laser spectral absorption measurement

Fig. 3. Machine-learning-based reconstruction model for temperature and concentration fields retrieval

Fig. 4. Numerical simulation of CH₄-air laminar coaxial diffusion flame. (a) Computation grid and boundary conditions for simulating flame; (b) temperature field; (c) CO₂ concentration field

Fig. 5. True flame temperature and CO₂ concentration fields, as well as machine-learning-based model predicted fields with 2%, 5% and 10% Gaussian random noises. (a) Temperature fields; (b) CO₂ concentration fields

Fig. 6. Comparisons of true temperature and CO₂ concentration fields (ideal) with machine-learning-based model predicted ones (MLP) with different random noises. (a) 2% random noise; (b) 5% random noise; (c) 10% random noise

Fig. 7. Comparisons of true temperature and CO₂ concentration distributions (ideal) at the height above burner of 40 mm with machine-learning-based model predicted ones (MLP). (a) 2% random noise; (b) 5% random noise; (c) 10% random noise