Reaserch On Intelligent Optimization Method For Radiation Shielding Design Of Small Compact Reactors

In recent years,due to the advantages of low site selection requirements,flexible applicability and low nuclear safety risk,the design and construction of small compact reactors has become an important direction of the development of national nuclear facilities,and its development has gradually become a research hotspot at home and abroad.In nuclear engineering design,especially for small compact reactors such as ship reactors and space reactors,the design optimization of shielding structure is not only an important part,but also a key link to ensure the economy and safety of reactors.However,due to the complexity of nuclear devices and the diversity of shielding design objectives,it is very necessary to realize an intelligent optimization method for radiation shielding design of small compact reactors on the premise of reducing the influence of uncertain factors such as human factors.Firstly,this paper verifies the feasibility of the intelligent optimization method of radiation shielding design based on neural network and genetic algorithm by using the plat shielding model,and the optimization results show that this method performs well in solving bi-objective optimization problems,but for the three-objective optimization problem with certain correlation between the optimization objectives,the optimization result has fell into the local optimization.Therefore,aiming at the above problem that the optimization results are not ideal due to mutual redundancy between optimization objectives,an intelligent optimization method for radiation shielding design based on neural network,principal component analysis and genetic algorithm is proposed,which can map the high-dimensional optimization objectives to several low-dimensional comprehensive components that are not related to each other in the optimization process,thus reducing the influence of redundancy between optimization objectives on optimization results.In addition,on the premise of constructing the radiation shielding calculation model of Savannah marine nuclear power reactor in equal proportion,aiming at different optimization problems,the above two different radiation shielding design optimization methods are used to optimize the shielding layer thickness parameters for different parts of the model,and the optimal solutions with uniform distribution in space is obtained,which verifies the feasibility of the two optimization methods in practical engineering.Finally,due to the complexity of the internal shielding structure of the reactors,on the premise of considering the irregular shielding structures such as the internal channels of the reactors,a radiation shielding calculation model of a small reactor is established.According to the thickness,material type and order of the calculation model and the shielding parameters such as the neutron flux on the outer surface after shielding,the neural network prediction model of the three-dimensional shielding structure is established.Based on the prediction model,the intelligent optimization methods of radiation shielding design are used to complete the optimization calculation for different optimization problems,and finally the optimization solution with uniform distribution in space is obtained.The research content of this paper gives the optimization results of two small reactor radiation shielding models in practical engineering applications,which can provide some theoretical support for reactor shielding structure designers,and is of great significance to the practical engineering design of small compact reactors.