船用反应堆屏蔽设计的可视化与快速计算功能开发

Visualization and Rapid Computing Function Development for Marine Reactor Shielding Design

船用反应堆的屏蔽设计问题直接关系到核能能否安全的用作舰船的动力系统。MCNP在船用反应堆的屏蔽计算中应用十分广泛,但其输入程序的编写及输出结果的整理较为繁琐,为了使用户更加简便的编写MCNP输入文件,直观的分析输出结果,本文开发了针对MCNP输入与输出文件的可视化软件。此外,在船用反应堆的屏蔽设计过程中需要MCNP进行大量屏蔽计算,所耗时间过长,为了实现在一定误差范围内的快速计算功能,本文采用BP神经网络模拟学习MCNP的计算过程,仅需给出指定的输入变量即可预测屏蔽计算输出结果,解决了MCNP计算耗时过长问题,提高了屏蔽设计优化效率。

Marine reactor shielding design problems directly related to nuclear power security as a ship power system.MCNP is widely used in Marine reactor shielding calculation,but writing the input files and analyzing the output files are complicated.In order for users more convenient to write the MCNP input files and intuitively analyze the output files,this paper developed a visualization software for MCNP input and output files.In addition,the process of Marine reactor shielding design need a lot of shielding calculation by MCNP,and it spent a long time.In order to achieve the fast computing functions within a certain error range,this paper uses the BP neural network simulate and study calculation process of MCNP,and only by giving the specified input variables,the net can predict the output of shielding calculation.It solved the problem that computation time is too long and raised the efficiency of shielding design optimization.