Scholarlink

With continuous development of modern science and technology, nuclear energy has become one of the most important energy sources in the world. The control rod drive mechanism (CRDM) is the only active device to control the operation of nuclear reactors and is the key equipment for reactor power regulation. As CRDM works in high temperature, high pressure and strong vibration environment, its performance degrades during long-term operation. Based on this, this paper proposes a method for predicting the remaining life of CRDM based on reconstruction of monitoring signals and adaptive boosting with convolutional neural network (Adaboost-CNN). Firstly, the complete ensemble empirical mode decomposition with adaptive noise (CEEMDAN) is used to reduce the noise of monitoring signals of CRDM during operation, and the current signal is corrected by historical data to obtain the reconstructed signal. Secondly, a weak prediction model for the remaining life of CRDM is established based on convolutional neural network (CNN). After that, a strong prediction model can be established based on multiple weak prediction models and Adaboost. Finally, the proposed intelligent remaining life prediction method is verified. The experimental results have shown that the proposed method can accurately predict the remaining life of CRDM and is superior to traditional methods.

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 Remaining Life Prediction of the Reactor CRDM Based on Monitoring Signal Reconstruction and Adaboost-CNN

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Zhi-long Liu,
  • Zhen Zeng,
  • Qin-chao Wang,
  • Yan-ping Huang,
  • Chang-hua Nie,
  • Tian-zhou Xie

摘要

With continuous development of modern science and technology, nuclear energy has become one of the most important energy sources in the world. The control rod drive mechanism (CRDM) is the only active device to control the operation of nuclear reactors and is the key equipment for reactor power regulation. As CRDM works in high temperature, high pressure and strong vibration environment, its performance degrades during long-term operation. Based on this, this paper proposes a method for predicting the remaining life of CRDM based on reconstruction of monitoring signals and adaptive boosting with convolutional neural network (Adaboost-CNN). Firstly, the complete ensemble empirical mode decomposition with adaptive noise (CEEMDAN) is used to reduce the noise of monitoring signals of CRDM during operation, and the current signal is corrected by historical data to obtain the reconstructed signal. Secondly, a weak prediction model for the remaining life of CRDM is established based on convolutional neural network (CNN). After that, a strong prediction model can be established based on multiple weak prediction models and Adaboost. Finally, the proposed intelligent remaining life prediction method is verified. The experimental results have shown that the proposed method can accurately predict the remaining life of CRDM and is superior to traditional methods.