Study On Optimization For Radiation Symmetry Of Indirect Drive Based On Compressed Sensing

The key to realize controlled nuclear fusion ignition is to reduce the radiant-drivenasymmetry on target pellet in Inertial Confinement Fusion (ICF). Evaluate the symmetry ofradiant energy on the surface of target pellet by simulating the process of target pelletimplosion indirectly driven by laser radiation. We can optimize the structure of hohlraum andimprove the success rate of the experiments. However, the calculation efficiency of thesimulation is subject to the complexity of the hohlraum structure and the amount of numericalcalculation. In this thesis, systematic study on how to improve the efficiency of optimaldesign of drive symmetry on target pellet is discussed.The simulation-based optimization for drive symmetry of the target pellet is based on theanalysis of the calculation of the radiant-driven symmetry. On the basis of the model ofenergy balance, divide the radiation surface of the holhraum and the surface of target pelletinto three-dimensional mesh, and calculate the high energy distribution on target pelletthrough discrete view factor method. After that, calculate the drive symmetry of the targetpellet with spherical harmonic decomposition.The simulation-based optimization of drive symmetry of the target is to calculate thedrive symmetry of the target pellet on different hohlraum parameters, getting the responsesurface and find out the optimal value. As the discrete number of the design variables and themesh number of the hohlraum and the target pellet increase, the amount of numericalcalculation increase hugely, this leads to the extension of the calculation time for the responsesurface of the drive symmetry of the target pellet in simulation process, and restricts theapplication of the software. In the meanwhile, the commonly used response surfacemethodology cannot solve the problem due to the uncertainty of the test results of thesimulation-based optimization.The reconstruction of response surface based on the compressed sensing theory is raisedin this thesis targeting the problem. The thesis is focused on reconstruction of responsesurface. Base on the sparse representation of response surface, calculate the corresponding response value by non-adaptive sampling. Then reconstruct the response surface by nonlinearconjugate gradient method. The efficiency, precision and stability of the method are alsoevaluated. Simulations are performed on some typical radiation models, comparisonbetween the simulation for optimization of response surface based on the compressed sensingtheory and traditional optimization of response surface shows the advantage of the newmethod. The new theory method improves the stability with the guarantee of precision, andcan get a better optimization result.Finally, an analytical and visualization software for radiant-driven symmetry analysisbased on the compressed sensing method is introduced. It’s mainly about the frame and thefunctional modules; a typical example about the optimization is also given.