Scholarlink

A prompt γ neutron activation analysis (PGNAA) detection system was established. The 14 MeV D-T neutron generator developed by Northeast Normal University was used as the neutron source, and the BGO detector was used to collect the γ energy spectrum of a 2 kg graphite sample. The denoising effects of the EMD method and the wavelet soft and hard threshold methods on the energy spectrum were compared. It is proposed to use the IMF of the γ energy spectrum as high frequency noise for the calculation of the wavelet denoising threshold function, and then denoise the energy spectrum. The denoising effect of the algorithm was evaluated by parameters such as smoothness (S), signal to noise ratio (SNR), and root mean square error (RMSE). Using the optimized wavelet threshold calculation method, the hard threshold denoising with 5 decomposition layers and the sym4 basis function achieved the best effect. Compared with the traditional method, the signal to noise ratio of the energy spectrum was improved by 11.2%, and the root mean square error was reduced by 34.7%.

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 Optimization of Wavelet Denoising Threshold Function in PGNAA Based on Empirical Mode Decomposition Technology

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Hailong Xu,
  • Ke Gong,
  • Yingying Cao,
  • Pingwei Sun,
  • Shangrui Jiang,
  • Jiayu Li,
  • Sijia Zhou,
  • Jia Song,
  • Shengduo Liu,
  • Weiyang Zhang,
  • Siqi Liu,
  • Zebin Li,
  • Yuxuan Gu,
  • Yue Sun,
  • Shiwei Jing

摘要

A prompt γ neutron activation analysis (PGNAA) detection system was established. The 14 MeV D-T neutron generator developed by Northeast Normal University was used as the neutron source, and the BGO detector was used to collect the γ energy spectrum of a 2 kg graphite sample. The denoising effects of the EMD method and the wavelet soft and hard threshold methods on the energy spectrum were compared. It is proposed to use the IMF of the γ energy spectrum as high frequency noise for the calculation of the wavelet denoising threshold function, and then denoise the energy spectrum. The denoising effect of the algorithm was evaluated by parameters such as smoothness (S), signal to noise ratio (SNR), and root mean square error (RMSE). Using the optimized wavelet threshold calculation method, the hard threshold denoising with 5 decomposition layers and the sym4 basis function achieved the best effect. Compared with the traditional method, the signal to noise ratio of the energy spectrum was improved by 11.2%, and the root mean square error was reduced by 34.7%.