节点文献
太赫兹时域光谱结合偏最小二乘对不同沙源地样品分类的研究
Classification of sand samples by using terahertz time-domain spectroscopy and partial least squares models
【作者】 刘盼盼; 张卓勇; 潘宝林; 魏明建; 朱若华; 张欣;
【Author】 Panpan Liu;Zhuoyong Zhang;Baolin Pan;Mingjian Wei;Ruohua Zhu;Xin Zhang;Department of Chemistry,Capital Normal University;Department of Resource and Tourism,Capital Normal University;
【机构】 首都师范大学化学系; 首都师范大学资源环境与旅游学院;
【摘要】 我们提出了一种利用透射太赫兹时域光谱与化学计量学中偏最小二乘分类建模相结合的方法对取自不同沙漠中的不同粒径的砂样进行了分类研究。本研究中我们把测得的透射太赫兹时域光谱吸收光谱,首先经过移动窗口拟合多项式求导和直接正交信号校正法进行预处理,然后采用偏最小二乘的方法对处理后光谱建立分类模型,结果表明,该模型对砂样粒径和来源地分类的预测正确率在85-100%。研究中,我们还采用了扫描式电子显微镜来观测砂样颗粒表面的形貌差异,用能谱仪测定了砂样中主要元素含量来探讨砂样能够正确分类的原因。结果表明,太赫兹时域光谱与多元建模方法相结合,对确定中国北方地区形成沙尘暴的砂样来源工作提供可靠的信息,对进一步研究沙尘暴形成和发展机制具有很大的意义;同时,这项研究表明了太赫兹时域光谱可以对不同沙漠砂样的来源进行鉴定,并对于自然地理研究中探究沉积物的来源是提供了一项有用的工具。
【Abstract】 A novel method combining partial least squares(PLS) with terahertz time domain spectroscopy(THz-TDS) was proposed for identification of sand samples from different sand lands and in different grain sizes.The spectra were collected and then Savitzky-Golay and direct orthogonal signal correction were used as pretreatment method,and finally PLS method was used to establish classification models.The predicted correct classifications between 85-100%were obtained.Scanning electron microscopy was used to investigate the differences of morphology.The sand samples were analyzed by using energy dispersive spectroscopy and principal elemental contents were obtained.Results showed that THz-TDS combined with multivariate modeling methods can provide reliable and useful information to identify the sources of sandstorms formed in northern China region.This work reveals that THz-TDS can be a useful tool for identification of sand samples from different sand lands and for tracing sources of the sediments from the natural geographical studies.
- 【会议录名称】 中国化学会第30届学术年会摘要集-第二十五分会:化学信息学与化学计量学
- 【会议名称】中国化学会第30届学术年会-第二十五分会:化学信息学与化学计量学
- 【会议时间】2016-07-01
- 【会议地点】中国辽宁大连
- 【分类号】P425.55
- 【主办单位】中国化学会