摘要
基于统一计算架构(Compute unified device architecture,CUDA),利用GPU的流多处理器和共享显存实现了辐射环境中人体体素级受照剂量的并行计算。通过加速器驱动次临界系统散裂靶质子束窗的维修更换仿真例题的测试,百万量级体素人体受照剂量计算时间降低到10 ms的量级,能够保持与实时维修仿真同步。该方法能够满足维修仿真对维修人员的器官当量剂量和人体有效剂量计算的实时性要求,可应用到维修方案的实时仿真推演,对提高维修方案的设计与验证效率,辅助优化方案和保障人员辐射安全具有重要意义。
The absorbed dose assessment method was developed based on compute unified device architecture parallel architecture and human voxel model, to realize the parallel computing of radiation dose for human body in the radiation environment using the GPU processors and shared memory. The maintenance strategy simulation dose assessment of the spallation target proton beam window for accelerator driven system was tested by this method. The results showed that the computing time can reach 10 ms level with millions of voxel and the dose calculation can synchronize with real-time simulation of maintenance process. This method could realize the requirement of organ equivalent dose and effective dose calculation for maintenance worker. In the simulation deduction of maintenance strategy, this method is of great significance in protecting personnel radiation safety and improving the efficiency forthe design and validation of maintenance scheme.
出处
《辐射研究与辐射工艺学报》
CAS
CSCD
2015年第6期41-47,共7页
Journal of Radiation Research and Radiation Processing
基金
中国科学院战略性先导科技专项(XDA03040000)
国家ITER973计划(2014GB1120000)
国家自然科学基金(No.11305205)资助~~
关键词
受照剂量评估
并行计算
统一计算架构
实时仿真
Absorbed dose assessment, Parallel computing, Compute unified device architecture, Real-timesimulation