Reasearch On Combustion Diagnosis And Reconstruction Method Based On Flame Emission Tomography (FET)

The combustion diagnostic plays important roles in energy engineering,transportation,and aerospace industries,which has great potential in combustion efficiency improvement and polluting gases control.The three-dimensional(3D)visualization of the combustion field and the measurement of key physical parameters such as temperature,species concentration and velocity during the combustion process are important topics in the field of combustion diagnostic.Benefiting by the noncontact and nonintrusive advantages of the optical detection method and the advantages of the 3D full-field measurement of the measured field by computational tomography,flame emission tomography(FET)has ability to realize nonintrusive and instantaneous 3D quantitative measurement and 3D full-field visualization of key physical parameters in the combustion field,which has very important research significance in combustion diagnostic.This Ph.D thesis mainly focus on the basic principles and reconstruction algorithms of FET,and proposed a series of 3D reconstruction algorithms and its applications on account of the problems of camera overexposure,the simultaneous dynamic measurement of multiple combustion chemiluminescence,the limited acquisition direction of the tomography system and high efficient reconstruction.The main works are summarized as follows.Research on the calculation model of weight coefficient of FET.Based on the existing lens imaging model,the positional relationship between imaging pixels and blurry circle is refined,and improved calculation models of flame emission tomography are proposed to make it more appropriate to the practical imaging process.Both numerical simulating estimations and propane flame experiments illustrate the feasibility of the improved calculation model in combustion diagnostic.Hybrid algorithm for imaging overexposure of FET.In order to compensate imaging overexposure during combustion process,a hybrid algorithm combining weight correction and Tikhonov's regularization is proposed.The numerical simulation quantitatively evaluates the performance of the hybrid algorithm,and the reconstruction ability of the projection with different saturation ratios is explored.Additionally,an experiment system consisting of multiple cameras was established to reconstruct the 3D combustion structure of propane flame with different exposure time settings based on algebraic reconstruction technique(ART)algorithm and the proposed hybrid algorithm.The obtained results show that the hybrid algorithm can effectively compensate for the loss of flame peak information caused by overexposure and improve the reconstruction quality.Regularization reconstruction algorithm for FET.Combined with the traditional iterative algorithm and the prior knowledge of the combustion field,a 3D reconstruction algorithm based on total variational sparse regularization is proposed.Through numerical experimental analysis,it is proved that in the experiments in this paper,the proposed algorithm's reconstruction error is only 50% of the ART algorithm.A method of selecting regularization parameters based on projection error is discussed.Finally,a qualitative and quantitative comparison of 3D reconstruction results of candle flames based on this algorithm and ART algorithm was performed,which proved that the algorithm can improve the inevitable artifacts existing in ART algorithm reconstruction results,and achieved the purpose of improving reconstruction quality through multi-information fusion.Research on multi-component FET reconstruction technology.A color separation method based on Bayer format image for flame species concentration of CH* and C2* is proposed.The dual-wavelength solid laser is used as the simulation light source to verify the accuracy of the algorithm and figure out that the average root mean square error is less than 0.5%.Flame emission tomography combining with multi-directional simultaneous capturing is proposed for real time three dimensional observations and detection in flame,which achieve the simultaneous dynamic measurement and 3D visualization of CH* and C2* of instantaneous non-axisymmetric paopane flame.The proposed method is helpful to improve the simultaneous acquisition and quantitative separation of multiple emission spectra in combustion diagnosis.which has potential in the measurement and diagnosis of key physical parameters such as temperature,equivalent ratio and velocity of the combustion field in the future.FET reconstruction via deep learning.Combining deep learning and 3D reconstruction,a3 D flame field fast reconstruction technology based on convolutional neural network(CNN)is proposed.Based on this method,the three-dimensional CH* and C2* structure reconstruction of three types of candle flames was performed,which verified the feasibility of directly using CNN instead of CT reconstruction algorithm for three-dimensional reconstruction.Through comparison with the 3D reconstruction results of other CT reconstruction algorithms,it is found that in the experiment of this paper,the CNN-based reconstruction algorithm takes only 1% of the calculation time of the existing iterative reconstruction algorithm,which strongly proves the superiority of the proposed algorithm in calculation time.Based on the trained CNN,real-time reconstruction of the combustion field can be achieved.In addition,the 3D fast reconstruction based on the CNN network model under different camera exposure conditions is discussed.Furthermore,the feasibility of 3D FET rapid reconstruction technology based on CNN is estimated with limited total projection angle.