Research On TGS Image Reconstruction Technology Of Nuclear Waste Bin Based On Compressed Sensing Theory

The information of material and radionuclide are important basis for effective discrimination and classification of radioactive waste drum.The radioactive waste drum must be assayed in the nuclear safety monitoring and radioactive waste treatment and disposal.Tomographic gamma scanning(TGS)is one of the most advanced non-destructive techniques for assaying the radioactive waste drum.The images of linear attenuation coefficient and radionuclide activity are reconstructed by scanning the radioactive waste drum with the gamma detector,which achieves the qualitative,quantitative and positional analysis of radionuclide in the drum.Existing TGS systems are difficult to have high assay efficiency,better radionuclide identification ability and imaging quality,which seriously restrict the practical application of these systems.The rapid assay and high-quality imaging for radioactive waste drum are technical difficulties of TGS.This work applies the compressed sensing method to assay the radioactive waste drum and reconstruct TGS image.The radionuclide is accurately identified by a gamma spectrometer with an HPGe detector.Combined sparse measurement of gamma spectrum and optimized reconstruction algorithm,the assay efficiency and image reconstruction quality of TGS system are improved.This work researches the compressed sensing TGS image reconstruction theory,radioactive waste drum TGS scanning experiment,efficiency calibration method,image reconstruction algorithm and image reconstruction analysis software.The contents and conclusions are as follows:(1)A method for assaying the radioactive waste drum and reconstructing TGS image with a single HPGe based on compressed sensing is proposed.This method ensures better radionuclide identification ability and improves the assay efficiency and image reconstruction quality.Seven different materials and a standard point source 137Cs are used to construct a heterogeneous radioactive waste drum sample.Taking the center of the drum as the origin point in polar coordinates,a TGS image is divided into 864(12 circles×72 angles)voxels.96 projection data are measured by experimental scanning with 4 horizontal and 24 angular positions with the sparse measurement,which improves the efficiency of TGS scanning.The binomial relationship between projection data and relaxation factor is established to improve the ART iteration and IART-TVM algorithm is implemented with the image total variation minimization iteration and applied to reconstruct the TGS image.For images of linear attenuation coefficient of six energies,the mean square errors of IART-TVM algorithm are about 0.64 times that of ART algorithm and the signal-to-noise ratios of IART-TVM algorithm are about 1.64 times that of ART algorithm.The positioning accuracy of 137Cs in the image of activity reconstructed by IART-TVM algorithm is much better than that of ART algorithm.The reconstructed activity of 137Cs is 3.378×105Bq and the relative deviation is 3.21%.The IART-TVM algorithm greatly improves the accuracy of the reconstructed activity.(2)The MLEM algorithm is applied to reconstruct the original image with high fidelity quality.MLEM-TVM optimization algorithm is implemented to optimize the whole iterative process of compressed sensing image reconstruction.Four different materials and a standard point source 137Cs are used to construct a heterogeneous radioactive waste drum salple.Nuclide recognition and projection data extraction are accurately accomplished by a gamma spectrometer system with an HPGe detector based on sparse measurement.MLEM.IART-TVM and MLEM-TVM algorithls are applied to reconstruct the TGS image.For images of linear attenuation coefficient of six energies,the mean square errors of MLEM-TVM algorithm are about 0.4 times that of MLEM algorithm and 0.74 times that of IART-TVM algorithm.The signal-to-noise ratios of MLEM-TVM algorithm are about 2.34 times that of MLEM algorithm and 1.24 times that of IART-TVM algorithm.The activity of 137Cs reconstructed by MLEM-TVM algorithm is 3.152×105Bq and the relative deviation is 3.01%.The activity of 137Cs reconstructed by IART-TVM algorithm is 3.193×105Bq and the relative deviation is 4.35%.The uncertainties of reconstructed 137Cs activity are less than 5%by iterative algorithms with TVM iteration of compressed sensing.Compared with MLEM algorithm,MLEM-TVM and IART-TVM algorithms have better reconstruction effect.The quality of initial image obtained by MLEM algorithm is higher than other algorithms.Therefore,MLEM-TVM algorithm is more suitable for TGS image reconstruction with the under-sampling condition of projection data because of its superior performance.(3)An optimization method of TGS efficiency calibration based on point source spatial efficiency function is proposed.This method solves the problems of time lag,poor generality and heavy workload of existing methods.The MCNP model of the TGS detection system is established to calculate the detection efficiency of different point source locations and gamma energies.Combined with proposed the point source spatial efficiency function model,those point source efficiencies are fitting to construct the efficiency calibration piecewise function and parameters by using multivariate nonlinear regression.Point source efficiencies of 241 Am and 133Ba are simulated by MCNP and that of 137Cs and 60Co are experimental measured at 507 points of the TGS system detection space.Correlation coefficients R2 of efficiency measurement and calibration function calculation are close to 1 for gamma rays of 0.059MeV,0.081MeV,0.662MeV,1.173MeV and 1.332MeV,which verifies the accuracy of point source spatial efficiency calibration function.Because the size of voxel in the TGS image is very small,the voxel efficiency is approximately equivalent to the point source efficiency of the voxel center.When the voxel central position coordinate and gamma energy of 0.662MeV are brought into the efficiency calibration function,the spatial distribution image of voxel efficiency at four detection positions of detector are obtained and die voxel efficiency matrix of every scanning state is calculated quickly and accUrately.(4)A TGS image reconstruction software for radioactive waste drum is designed with MATLAB GUI to analyze the gamma spectrum,reconstruct the images of linear attenuation coefficient and radionuclide activity in drum and calculate the total activity.The gamma spectrum analysis method,TGS efficiency calibration optimization method and MLEM-TVM optimization algorithm are applied to this software.Compared with the analysis results of MAESTRO software,for 96 transmission gamma spectra,the relative deviations of net peak area of this software are less than 6%.For 96 emission gamma spectra,the relative deviations of net peak area are less than 2%.The total activity of 137Cs reconstructed by this software is 3.143×105Bq and the relative deviation is 2.71%.The software achieves the function of TGS image reconstruction for radioactive waste drum and the uncertainty of analysis meets the requirement of gamma spectrum analysis and TGS image reconstruction.This work researches the compressed sensing TGS image reconstruction theory,radioactive waste drum TGS scanning experiment,efficiency calibration method,image reconstruction algorithm and image reconstruction analysis software.A method for assaying the radioactive waste drum and reconstructing TGS image with a single HPGe based on compressed sensing is proposed.An optimization method of TGS efficiency calibration is established.A TGS image reconstruction software is designed.This work has completed the accurate recognition of radionuclide and improved the assay efficiency and image reconstruction quality for TGS system with a single HPGe detector and it has reference value.