Inverse Problem For Acoustic Computer Tomography And Reconstruction Of Temperature Field

Temperature field measurement of acoustic computer tomography calculates temperaturedistributions of the measurement region based on multi-path acoustic travel-times data, whichis classified as the "reverse results into effects" study of inverse problems, has the advantagesof non-contact without interfering temperature fields, wide temperature range (0?~2000?),large range of measured object space (dozens of meters) and on-line measurement, etc.Acoustic pyrometer is one of the typical applications of the technique in the furnacetemperature field monitoring. While temperature distribution monitors of atmosphere anddeep-sea hydrothermal and stored grain are new application areas of acoustic computertomography in exploration. In this paper, solution of inverse problem for the acousticcomputer tomography, reconstruction algorithm of temperature field and other issues areanalyzed theoretically and experimental studied deeply, The main work is completed asfollows:Major causes that affects acoustic computer tomography measurement precision intemperature field are analyzed in this paper. Aiming at those existing problems in acousticcomputer tomography reconstruction algorithm on temperature distribution, i. e.: the gridnumber of measured region division has to be less than number of effective sound wave pathswhich limits the original number of pixels, and results in low spatial resolution of thereconstructed temperature field, a3D temperature field reconstruction algorithm based onMarkov radial basic function and Tikhonov regularization---3DMTR (Markov radial basicfunction and Tikhonov Regularization) is proposed. The number of partition grids in measuredarea can be much greater than number of sound wave paths by using this algorithm, which ismuch more suitable for complex temperature field reconstruction.The rationality of the sound Transceivers array design directly affects the real-timemeasurement of acoustic computer tomography experimental system and the accuracy oftemperature field reconstruction. In order to monitor temperature distribution in stored grainby the method of acoustic computer tomography, the parameters such as the number of soundtransceivers arranged around the cylindrical granary, the layout and the effective sound pathselection are designed by computer simulation. The temperature field reconstruction results of three typical three-dimensional model show that removal on the silo wall sound paths in thesame bus can reduce the temperature field reconstruction error. Increasing the number of thetransceivers on one layer or the number of the layers transceivers arranged, are able to reducethe reconstruction error. But setting a plethora of sound transceivers may increase thecondition number of the coefficient matrix, and then increase the reconstruction error. Throughthe simulation design of sound transceiver array, transceiver array parameters can beoptimized, and the performance of acoustic computer tomography experimental system can beimproved as well as the reconstruction precision of temperature field.Temperature field reconstruction by acoustic computer tomography is an ill-posedproblem. The choice of regularization parameter has an important influence on the accuracy ofreconstruction. The temperature field reconstruction algorithm of the2DMTR-A (MTR withAdaptive Regularization Parameter) and the3DMTR-A on measured area raised number ofpixels can be far more sound wave paths, selection of adaptive regularization parameter isproposed. They use a novel adaptive regularization parameter selection method, named asminimum change criterion, to determine a proper regularization parameter, which can make agood compromise between de-noise and detail reconstruction of temperature field. Thereconstruction results of simulation and experimental show that the regularization parametersselected by the minimum change criterion is superior to those selected by the common L-curvemethod because they produce lower reconstruction errors, with a wide range of adaptability.Therefore,2DMTR-A and3DMTR-A algorithm has good practicality.When sound wave propagates in un-uniform temperature fields, its path is bent due torefraction. Approximately sound wave paths as straight line will cause great errors, especiallyfor temperature field whose temperature gradient is steep. In order to improve thereconstruction accuracy of un-uniform temperature fields by acoustic computer tomography,the2D/3D reconstruction algorithm is proposed in which the bending of sound wave paths isconsidered. Firstly, reconstruct the2D/3D temperature field without considering the bendingof sound wave paths by using2D/3D reconstruction algorithm. Secondly, find the eigenrays inthe reconstruction field by using the triangular/tetrahedron forward deployment method forsound ray tracing after evaluating the emergent angle of eigenrays by usingshooting-interpolation method. Thirdly, establish the relationship between the acoustictravel-times over the eigenrays and the temperature distribution. Lastly, reconstruct the2D/3D temperature field with consideration of sound wave paths bending. The reconstruction resultsshow that the reconstruction accuracy can be improved obviously by considering the bendingof sound wave paths. The larger the temperature gradient of temperature field is the moreobviously the improvement effect.By adopting acoustic computer tomography experimental system based on virtualinstrument of LABVIEW, temperature field reconstruction algorithm proposed in this paperreconstructs the air2D/3D and stored soybean2D temperature fields, experimental resultverifies the effectiveness and reasonableness of the temperature field reconstruction algorithmpresented in this paper.