Research On Laser Absorption Spectroscopy Tomography Method Based On Bayesian Estimation

Combustion is the main source of energy in today’s society,and more than 60% of the world’s electricity is generated through combustion.However,the combustion process produces pollutants that have a significant impact on the environment.In order to reduce the production of pollutants,improve the efficiency of combustion and save energy,realtime monitoring of combustion processes is essential.Tunable diode laser absorption spectroscopy tomography(TDLAST)technology can reconstruct the distribution of flow field parameters such as concentration and temperature in the measured area and has a wide range of application prospects.In order to reconstruct the distribution of flow field parameters in the measured region more rapidly and accurately,this paper investigates the laser absorption spectroscopy tomography method based on Bayesian estimation with a priori information,and the main work is as follows:Firstly,based on the expected patch log-likelihood prior model of the flow field parameter distribution,the paper proposes a temperature imaging algorithm based on the expected patch log-likelihood prior.The scheme introduces the expected block loglikelihood prior into the solution of the TDLAST inverse problem and constructs a block prior model of the local absorption density based on a Gaussian mixture model.Compared with the existing TDLAST regularization algorithm based on the smoothed prior,this scheme can reconstruct the position and shape of the flame more accurately and clearly describe the temperature distribution within the region of interest of the combustion field.Secondly,in order to supplement the detailed information in the reconstructed temperature distribution images,a two-stage temperature imaging algorithm based on Tikhonov regularization is designed in this paper.The scheme is divided into two stages,with the first stage reconstructing the local absorption density of the gas and the second stage complementing the reconstructed results of the first stage.Simulation experiments and actual combustion field experiments show that the scheme not only improves the accuracy of the reconstructed flame position and shape,but also complements the detailed information contained in the reconstructed temperature images.Finally,in order to further improve the reconstruction quality of the overall temperature image,a two-stage temperature imaging scheme based on the expected patch log-likelihood prior is designed in this paper.The scheme establishes block prior models for local absorption density and residual local absorption density using Gaussian mixture models with different block sizes and model numbers in the initial reconstruction stage and detail supplement stage,respectively,and solves the TDLAST inverse problem with the introduction of Gaussian mixture model regularization using the alternating direction multiplier method.This scheme not only improves the reconstruction accuracy of flame temperature in the high temperature region,but also further improves the accuracy of detail feature reconstruction in the flame thermal radiation region and low temperature region.