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

As a kind of chemical reaction,the process of combustion is very complex,including a variety of chain reactions between different groups.Therefore,it is very important for the safe operation and energy saving of the combustion furnaces to have an on-line monitoring of the temperature distribution and some chemical products.Soot is a very typical solid product which is produced from the process of combustion.Its particle size is usually below 0.1 ?m and it can be easily absorbed from the body's respiratory system into the lungs,affecting people's health seriously.Using the LCTF-based hyperspectral flame imaging system,a three-dimensional model was established for the reconstruction of the temperature and soot volume fraction field.A lot of numerical simulations and experiments have been done in the paper.The main contents and conclusions are as follows:First,the radiative transfer equation in the flame was deduced.According to the HITRAN database,the emission spectra of several typical gas products were imquired.It can be found that the radiation of gas products in the flame can be neglected in the working wavelength band of 650 to 1100nm.The MIE calculation program was used to calculate the radiation characteristics of the soot,carbon particles and ashes in the range of 650-1 100nm.It was concluded that the scattering of soot can be neglected and it is appropriate to consider the absorption of soot only.On the contrary,in terms of ashes,it is appropriate to consider the scattering of ashes only and the absorption process can be neglected.For carbon particles,the absorption and scattering should be taken into account together,but the scattering coefficient is slightly larger than the absorption coefficient.The definition of radiative inverse problem was explained here.Three different calculation methods,including Tikhonov regularization algorithm,truncated singular value decomposition(TSVD)method and least squares QR decomposition(LSQR)method,were used to calculate the three dimensional temperature distribution by solving the radiative inverse problem.And the LSQR algorithm was considered to be more suitable solving the complex radiative inverse problem in flames than the other two algorithms,considering both reconstruction error and reconstruction time.A reconstruction model was proposed here for the reconstruction of 3-D temperature and soot volume fraction distributions in the hydrocarbon flame by the multi-wavelength emission method.The multi-wavelength method and the original two-color method were both used in the case of symmetrical flame under different SNRs.The comparions results showed that when the SNR is 54dB,the average error of temperature reconstruction was 0.4806%,and the average error of soot volume fraction was 8.003%using the two-color method;while using the multi-wavelength method,the average error of temperature reconstruction was 0.2401%,and the average error of soot volume fraction was 3.17%.It was clear that the multi-wavelength emission method has a stronger noise resistance than the two-color method.Considering both the reconstruction error and computational efficiency,the optimal number of wavelengths for multi-wavelength emission method was determined to be 6-11.When dealing with asymmetric flames,the reconstruction results were considered to be satisfactory too.As a result,it was indicated that the proposed reconstruction model can be applied to both symmetric and asymmetric flames.A measureing model was proposed here to calculate the particle size distribution using the multi-wavelength emission method.To validate the feasibility of the model,numerical simulations about unimodal and bimodal paiticle size distributions were carried out using two different algorithms.It was found that the LSQR algorithm is more accurate in solving the linear equations with particle size distribution.After analyzing the correlation between the equations,a same-interval wavelength chosing method was proposed to chose the wavelengths used in the calculation.And the most optimal number of the chosen wavelengths was considered to be 5.A laboratory-scale ethylene laminar diffusion flame was built using the hyperspectral flame imaging system based on LCTF.A standard blackbody furnace was used for the calibration of the hyperspectral CCD camera in the wavelength range of 650 nm to 950 nm.For the object of a symmetrical ethylene diffusion flame,the 3-D reconstruction of temperature and soot volume fraction distribusions were carried out using flame images from only one direction.The reconstruction results were compared with that measured from thermocouple and laser extinction method and they were considered to be in great agreement with each other.While for the object of a asymmetrical ethylene diffusion flame,the experimental reconstruction was carried out using the flame images from four different directions.The shapes of the reconstructed temperature and soot volume fraction distributions in the vertical cross-section were considered to be similar with the flame's asymmetric shape.In addition,the trend of the temperature and soot volume fraction in the vertical direction was proved to be consistent with that in the symmetric flame,which validated the rationality of the reconstruction results.