Scholarlink

<p>Thorium (Th), a promising nuclear fuel, requires accurate detection for efficient resource utilization. This study employed LIBS combined with PLSR and RF models to quantify Th in complex ores. Training set augmentation significantly improved prediction accuracy, reducing RMSEP to 495.58&#xa0;ppm (PLSR) and 306.00&#xa0;ppm (RF). PCA-based dimensionality reduction followed by SVM and RF classification yielded accuracies of 97.92% and 87.5%, respectively. Results highlight the effectiveness of training set optimization and hybrid modeling in enhancing LIBS performance. The proposed approach enables rapid, in situ Th analysis, offering strong support for thorium resource exploration and nuclear energy development.</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

Laser-induced breakdown spectroscopy combined with machine learning for thorium element analysis in ores

  • 期刊论文

  • 发表日期：

    2025年12月5日

查看全文
  • Shuqin Zhan,
  • Xizhu Wang,
  • Lingling Peng,
  • Min Zhang,
  • Wenmei Jiang,
  • Xiangfeng Liu,
  • Xiaolong Zhang,
  • Xiaoliang Liu,
  • Shaoxing Liu

摘要

Thorium (Th), a promising nuclear fuel, requires accurate detection for efficient resource utilization. This study employed LIBS combined with PLSR and RF models to quantify Th in complex ores. Training set augmentation significantly improved prediction accuracy, reducing RMSEP to 495.58 ppm (PLSR) and 306.00 ppm (RF). PCA-based dimensionality reduction followed by SVM and RF classification yielded accuracies of 97.92% and 87.5%, respectively. Results highlight the effectiveness of training set optimization and hybrid modeling in enhancing LIBS performance. The proposed approach enables rapid, in situ Th analysis, offering strong support for thorium resource exploration and nuclear energy development.