Scholarlink

<p>Chlorine-induced corrosion is one of the primary causes of mechanical deterioration in cement concrete structures. This study presents a comparative evaluation of the performance of commonly employed neutron sources (<sup>241</sup>Am-Be, <sup>252</sup>Cf, DD, and DT). Consequently, the DD neutron generator demonstrated superior performance, exhibiting a normalized sensitivity and a signal-to-noise (SNR) ratio at a 10&#xa0;cm corrosion depth that were 35.5 and 70.5% higher than those of the <sup>252</sup>Cf neutron source. Furthermore, a multilayer perceptron (MLP) deep learning model was developed to establish the mapping between the chlorine concentration distribution and the prompt γ-ray intensities from the (n,γ) reaction using a DD neutron generator. The prediction accuracy, as represented by the root mean square error (RMSE), mean absolute error (MAE), and coefficient of determination (R<sup>2</sup>), was 0.068, 0.013, and 0.999, respectively.</p>

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

Chlorine concentration detection in concrete using PGNAA: system optimization with NSGA-II and concentration prediction with multilayer perceptron

  • 期刊论文

  • 发表日期：

    2025年12月12日

查看全文
  • Lucheng Yang,
  • Mingfei Yan,
  • Xubin Zhang,
  • Jingru Chen,
  • Zhanfei Liu

摘要

Chlorine-induced corrosion is one of the primary causes of mechanical deterioration in cement concrete structures. This study presents a comparative evaluation of the performance of commonly employed neutron sources (241Am-Be, 252Cf, DD, and DT). Consequently, the DD neutron generator demonstrated superior performance, exhibiting a normalized sensitivity and a signal-to-noise (SNR) ratio at a 10 cm corrosion depth that were 35.5 and 70.5% higher than those of the 252Cf neutron source. Furthermore, a multilayer perceptron (MLP) deep learning model was developed to establish the mapping between the chlorine concentration distribution and the prompt γ-ray intensities from the (n,γ) reaction using a DD neutron generator. The prediction accuracy, as represented by the root mean square error (RMSE), mean absolute error (MAE), and coefficient of determination (R2), was 0.068, 0.013, and 0.999, respectively.