Scholarlink

The power distribution of High Flux Engineering Test Reactor (HFETR) varies relatively rapidly during operation. To ensure the safe operation of the reactor, real-time monitoring of its power distribution is essential. However, HFETR has not installed an in-core monitoring system and only has a limited number of detectors. Research on core power distribution reconstruction methods based on these limited measurements can enable real-time monitoring of the core power distribution of HFETR. In this paper, a core power distribution reconstruction model for HFETR is developed based on a Backpropagation (BP) neural network. The accuracy of the reconstruction model is further improved through optimization of hyperparameter configurations. The feasibility of applying neural network methods to reconstruct the core power distribution of HFETR is systematically explored. The results indicate that neural network can effectively reconstruct the core power distribution of HFETR, providing important data support for subsequent research, thereby enhancing the safe operation of research reactors and laying a foundation for the artificial intelligence of research reactors.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Research on the Reconstruction Method of HFETR Core Power Distribution Based on BP Neural Network

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Yutong Zou,
  • Liqing Qiu,
  • Guohua Wang,
  • Chang Liu,
  • Yuhao He

摘要

The power distribution of High Flux Engineering Test Reactor (HFETR) varies relatively rapidly during operation. To ensure the safe operation of the reactor, real-time monitoring of its power distribution is essential. However, HFETR has not installed an in-core monitoring system and only has a limited number of detectors. Research on core power distribution reconstruction methods based on these limited measurements can enable real-time monitoring of the core power distribution of HFETR. In this paper, a core power distribution reconstruction model for HFETR is developed based on a Backpropagation (BP) neural network. The accuracy of the reconstruction model is further improved through optimization of hyperparameter configurations. The feasibility of applying neural network methods to reconstruct the core power distribution of HFETR is systematically explored. The results indicate that neural network can effectively reconstruct the core power distribution of HFETR, providing important data support for subsequent research, thereby enhancing the safe operation of research reactors and laying a foundation for the artificial intelligence of research reactors.