Three Dimensional Reconstruction Of Temperature,Soot Volume Fraction Distribution And Gas Volume Fraction Distribution In Flames Based On Hyperspectral Imaging System

Fossil fuels will continue to dominate the energy sources that humans obtain primarily in the coming decades.By burning fossil fuels,humans convert chemical energy from fossil fuels into heat.Chemical fuels for power generation,heating,transportation,and industrial use are important sources of air pollution.H₂O and CO₂in the flame can inhibit the generation of particles in the flame,and the temperature will greatly affect the formation of the particles.In order to improve reburning efficiency and reduce pollution,we need to measure the temperature,particle concentration and gas concentration in the flame at the same time.Based on the radiation characteristics of diffuse medium and gas,a flame hyperspectral imaging model is established based on the law of flame radiation heat transfer.The model can simultaneously measure the three-dimensional temperature field,particle concentration field and gas concentration field of the flame.At the same time,a large number of numerical reconstruction work and experimental verification were carried out on the model.In this paper,the radiation transfer equation of diffuse medium is introduced in detail.A model for solving the three-dimensional temperature field and particle concentration field based on the inverse Monte Carlo method and the two-step discrete coordinate method is discussed.This model can improve the computational efficiency compared with the traditional model.In order to achieve the detection of the gas concentration,a hyperspectral imaging model is established to obtain information on the spectral dimension of the flame,and a gas absorption peak（such as H₂O absorption peak of 1870 nm）is included in the measurement wavelength range.According to the continuity of the absorption of the particles and the gas absorption is selective,the absorption of the particles and the absorption of the gas are separated.Finally the particle concentration and the gas concentration are obtained.Next,the solution algorithm of the hyperspectral imaging model is discussed.The definition of the ill-conditioned problem is given and it is pointed out that the solution of the hyperspectral imaging model is an ill-conditioned problem.The LSQR algorithm,the damped LSQR algorithm,the Tikhonov regularization algorithm and the truncated singular value decomposition（TSVD）algorithm are compared to solve the ill-conditioned equations.The damped LSQR algorithm is the optimal reconstruction algorithm considering both the reconstruction error and the reconstruction time.Most of the subsequent reconstruction work uses the damped LSQR algorithm.The radiation characteristics and gas radiation characteristics of the diffused medium are introduced in detail.The simplified calculation formula of the absorption coefficient of the specific diffuse medium is given.The absorption lines and absorption peaks of some common gases are given.The absorption intensity of the gas can be accurately calculated by the statistical narrow-band model and the line-by-line method.If the particle absorption coefficient or the gas absorption coefficient is known,the particle concentration and gas concentration can be inversely determined.Numerically solved the hyperspectral imaging model.Numerical simulation was used to analyze the effects of different signal-to-noise ratios,different soot concentrations and different coefficient matrices on the reconstruction results.Using an asymmetric flame,it was found that this measurement model can also reconstruct satisfactory results for it.Finally,the multi-wavelength emission spectrum reconstruction model is used.The numerical results show that the SNR of the H₂O concentration system measured by the multi-wavelength method can be as low as 54dB.The SNR of the H₂O concentration system measured by the three-wavelength method is at least 59 dB.Multi-wavelength methods can be applied in camera measurement systems with lower signal-to-noise ratios.The working principle and application of the hyperspectral flame imaging system based on liquid crystal tunable filter（LCTF）are introduced in detail.The flame reconstruction experiment platform,as well as the specific model,parameters and main functions of the instrument used in the experiment are introduced.The linear interpolation method is used to obtain the corresponding relationship between the gray value and the radiation intensity in the hyperspectral flame image.The ethylene flame was used as the experimental object,and the calculation results were compared with some of the measured results.Through the calculation of the flame hyperspectral image,the true image intensity of the flame image at 1870 nm is larger than that of the fitting calculation.The flame image of 1870 nm contains the gas absorption signal.Three-dimensional temperature field,soot concentration field and gas concentration field reconstruction experiments were performed on the flame images of 700 nm,900 nm and 950 nm using the three-wavelength method.Three-dimensional temperature field,soot concentration field and gas concentration field reconstruction experiments were performed on the flame images of 1200 nm,1300 nm,1400 nm,1500nm,1600 nm,1700 nm,1800 nm,1900 nm and 1870 nm using the multi-wavelength method.The reconstruction results are consistent with the general gas flame combustion characteristics.The accuracy of the reconstruction temperature was verified by thermocouple.The flame temperature,particle concentration and gas concentration distribution were verified by tunable laser extinction method.The measurement results were compared with the literature results.It was found that the hyperspectral imaging model can successfully reconstruct the flame three-dimensional temperature field,particle concentration field and gas concentration field.