摘要
本文介绍了新评价模型并定义了新目标函数,对求解模型所采用的模拟植物生长算法进行了改进,相关代码利用C++、CUDA编写并用GPU加速。利用两座大型γ辐照装置进行了排源实践,新剂量场分布采用不同的重铬酸银剂量计测量。结果表明:测量结果与理论计算结果符合较好,剂量不均匀度(DUR)相对偏差小于4.125%;新评价模型与原模型相比,DUR增加不明显,能量利用率提高了若干倍,计算时间从h量级缩短到min量级。检测结果验证了新方法的可行性和有效性。
A new evaluating model was introduced and a new target function was defined. The plant growth simulation algorithm (PGSA) solving the target function was improved.The related codes were written in C++ and CUDA,and were accelerated by GPU.Two big facilities from different cooperative companies were rearranged individually and their new dose field distributions were also detected individually by different potassium dichromate dosimeters.The results show that the experimental dose unniformity (DUR) accords well with the theoretical ones,and the relative deviation is less than 4.125%.Compared with the old evaluation model,the DUR obtained by new model increases little,the energy efficiency is improved for several times,and the computation time is reduced from hours to minutes.The results verify the feasibility and validity of the new method.
作者
杨磊
刘燕
罗志平
李文革
周毅吉
张美琴
YANG Lei;LIU Yan;LUO Zhiping;LI Wenge;ZHOU Yiji;ZHANG Meiqin(Yantai University,Yantai 264005,China;Hunan Academy of Agricultural Sciences,Changsha 410300,China)
出处
《原子能科学技术》
EI
CAS
CSCD
北大核心
2019年第5期932-938,共7页
Atomic Energy Science and Technology
基金
山东省自然科学基金资助项目(ZR2016AQ24)
关键词
Γ辐照
优化排列
多目标优化方法
GPU加速
γ irradiation
optimization deployment
multi-objective optimization me-thod
GPU acceleration
