Research On Deflection Tomography Method Of Combustion Temperature Field Based On Compressive Sensing

Combustion widely exists in life and industrial production.The research of combustion temperature field imaging method provides important data support for controlling pollutant emission,understanding combustion essence and improving combustion efficiency,which has important research significance and application value.Deflection tomography has the characteristics of non-interference,real-time,non-contact and so on.It has become a powerful tool for temperature field measurement.The temperature field measurement data is limited by the economic cost,optical path and measurement site.The traditional deflection tomography reconstruction has the problems of low resolution and poor accuracy under the condition of under sampling,compressed sensing is an effective method to solve the temperature field reconstruction under the condition of under-sampling.In this paper,we focus on the difficulty of selecting modulation factors in the reconstruction of compressed sensing temperature field,the sparse constraint does not satisfy the high-precision reconstruction,and the total variation constraint is not conducive to the description of the reconstruction details.The main research work is as follows:Aiming at the problem that the modulation factor in the compressed sensing deflection tomography total variation minimization algorithm cannot be adjusted adaptively based on experience,an adaptive adjustment factor total variation method is proposed.The adaptive modulation factor is based on the normalized norm of the difference between the calculated projection and the measured projection as an adaptive adjustment strategy to ensure that the modulation factor is large when the ₁l norm is large,otherwise,the modulation modulation factor is small,which solves the contradiction between the optimal solution of empirical selection of adjustment factor and the convergence speed.Aiming at the low utilization rate of prior information in the compressed sensing deflection tomography total variation method,a multi-directional total variation constrained reconstruction method is proposed.The use of multiple direction gradients to describe the prior information not only guarantees the information of the two gradient directions of horizontal and vertical but also makes good use of the angle information,which can more fully reflect the detailed information.The lack of directionality in total variation constraints is solved,and the utilization of prior information is improved.Aiming at the problem of large noise in total variation constrained reconstruction under the condition of under sampling,a joint constrained intelligent optimization reconstruction method is proposed.Dictionary learning has a strong ability to suppress noise.Therefore,in this paper,total variation and dictionary learning are combined to construct the reconstruction constraint framework,and the reconstruction efficiency is accelerated by the intelligent optimization method of longicorn beetle.In order to overcome the over smoothing problem in dictionary learning,a local weighted dictionary learning method is proposed based on the local total variation value of temperature field to represent the amount of detail information.If the total variation value is large,the local weight is small,otherwise,the local weight is large,thus solving the problem of over smoothing.In order to verify the feasibility and effectiveness of the theoretical method proposed in this paper,a multi-directional dynamic acquisition platform of Moirétechnology is built,and the projection data of temperature field under high under sampling conditions are obtained.On this basis,the temperature field is reconstructed by using the method proposed in this paper.Compared with the traditional method,the reconstructed temperature field isotherm retains more detailed information and improves the reconstruction accuracy,reconstruction speed and resolution.