基于人工智能技术的数字反应堆加速计算技术研究

打开文本图片集

中图分类号：TP18    文献标志码：A    文章编码：1672-7274（2025）08-0034-03

Research on Accelerated Computing Technology for Digital Reactors Based on Artificial Intelligence Technology

CHENGMing，WEI Zihao （China Academy of Atomic Energy，Beijing 102413，China）

Abstract：Digital reactor technologyisacuting-edgefieldof digital upgradingandresearchanddevelopment innovation in the fieldof nuclear reactorengineering.Currently，itfaces the dualchallengesofhigh-precision numerical calculationand rapid data exchange.The application of artificial intelligence technology to achieve highprecision data generation is a highly promising new method to solve this problem.This article conducts thorough researchandanalysis，selectingconditional generative adversarial neural network algorithms anddeep fullyconnected network algorithms basedon machine learning and deep learning.The shielding field calculationof smallfast neutron reactor istakenas teesearchoect，andartical intellgencemodel establshment，datagenerationandtraining，progam development，demonstrationcalculationandother workarecariedout.Comparedwith traditionalcomputationalanalysis programs，thecalculation speedcanbeimproved by4to5orders ofmagnitude within the samelmiliondata level and 3% error range.This proves that the research ideas and methods have significant application value and can further expand the application scope to support the development ofreal-time computing systems for digital reactors in the future.

Keywords：artificial intelligence; digital reactor; accelerated computation; shielding calculation

数字反应堆技术是核反应堆工程领域实现数字化升级、加速新型反应堆研发创新的重要手段，该技术可实现加速设计迭代、降低建造成本、试验数字替代等众多目标。（剩余3735字）