海洋学研究 ›› 2016, Vol. 34 ›› Issue (1): 18-26.DOI: 10.3969/j.issn.1001-909X.2016.01.003

• 研究论文 • 上一篇    下一篇

北部湾海底沉积物稀土元素与影响因子关系的BP神经网络定量分析

胡昊1, 许冬1, 龙江平*1,2, 周勐佳1, 唐博1, 金路1   

  1. 1.国家海洋局 第二海洋研究所,浙江 杭州 310012;
    2.浙江大学 地球科学学院,浙江 杭州 310007
  • 收稿日期:2014-11-18 修回日期:2014-12-03 出版日期:2016-03-15 发布日期:2022-11-24
  • 通讯作者: * 龙江平(1958-),男,研究员,主要从事海洋地球化学方面的研究。E-mail:hangzhow@126.com
  • 作者简介:胡昊(1990-),男,贵州印江县人,主要从事海洋地质科学方面的研究。E-mail:hhshiyi@gmail.com
  • 基金资助:
    国家自然科学基金项目资助(40776037);国家海洋公益性科研专项经费项目资助(201105003,201205008)

Quantitative analysis of BP neural network on the relationships between ∑REE content and impact factors in Beibu Gulf

HU Hao1, XU Dong1, LONG Jiang-ping*1,2, ZHOU Meng-jia1, TANG Bo1, JIN Lu1   

  1. 1. The Second Institute of Oceanography, SOA, Hangzhou 310012, China;
    2. School of Earth Sciences, Zhejiang University, Hangzhou 310007, China
  • Received:2014-11-18 Revised:2014-12-03 Online:2016-03-15 Published:2022-11-24

摘要: 海底沉积物中稀土元素的分布特征受多种影响因子的影响,很难定量分析。北部湾沉积物稀土元素(∑REE)与物源、水动力、沉积物粒度和黏土矿物百分比等关系定性分析显示,本区∑REE的物源主要由陆源岩石控制,弱水动力和细颗粒度都对应较高含量的∑REE。结合北部湾海底沉积物的位置、砾石含量、砂含量、粉砂含量、黏土含量和黏土矿物含量训练出来的BP神经网络在控制变量的情况下定量分析各影响因子与∑REE的关系,获得单个影响因子与∑REE的关系曲线。这些关系曲线揭示了北部湾沉积物中稀土元素与各影响因子的联系,所获结果与定性分析结果基本一致,该方法能够通过自主学习,自动判断并定量计算,有助于识别每一个因子对稀土元素含量影响的大小,是如何控制∑REE的分布,从而根据曲线的变化规律结合实际情况去推断区域的环境变化及地质演变,对稀土元素的富集和分散分析提供有益的理论指导。

关键词: 稀土元素, 影响因子, 定量分析, BP神经网络, 北部湾

Abstract: The distribution characteristics of ∑REE in the bottom sediments are influenced by many factors, so they are too difficult to analyze quantitatively. From qualitative analysis of the relationship between ∑REE content and its provenance, hydrodynamics, grain size and mineral distribution in Beibu Gulf, it is revealed that the main ∑REE composition is controlled by terrestrial rock. Both weaker hydrodynamics and the finer grain lead to the higher ∑REE content. Combined with the quantitative analysis of BP neural network which trained by the position of samples, gravel content, sand content, silt content, clay content and clay mineral content, the relative curves of individual impact factor with ∑REE content were achieved. These relative curves reveal the relationship between the ∑REE content and various impact factors, and the results are consistent with those of qualitative analysis. The method can study by itself, determine automatically and calculate quantitatively. It is helpful for identification of the impact related to each factor and ∑REE and to know how it is controlled the ∑REE distribution. So according to the curve variation and actual situation, environmental changes and geological evolution of the region can be inferred. This also provides a useful theoretical guidance for the analysis of the enrichment and dispersion for the rare earth elements.

Key words: REE, impact factors, quantitative analysis, BP neural network, Beibu Gulf

中图分类号: