摘要
使用科学而高效的方法检测天然岩石中矿物成分对于矿物的合理利用有重要的影响。文章使用近红外漫反射光谱仪获取由高岭土、白云母和蒙脱石3种岩石矿物粉末混合成的模拟天然岩石样本的近红外漫反射光谱信息,通过标准归一化的方法对光谱数据进行预处理,采用随机森林进行数学建模,对岩石样本的组成成分进行预测,预测得到3种岩石成分最小均方根误差分别为:0.0880,0.0956,0.1212。实验结果表明应用近红外漫反射光谱来测定天然岩石中各种矿物成分的含量是可行的,为今后岩石成分的快速检测提供了理论依据。
It is important to use scientific and efficient method to detect the content of mineral composition in natural rock for the ratio- nal use of mineral.In this paper,near-infrared diffuse reflectance spectrometer is used to obtain the near infrared diffuse reflection spectrum information of the simulate natural rock samples mixed by the rock powder of kaolin,white mica and montmorillonite.Standard normalization method is employed to pre-process the spectral data,Random forests is also adopted to establish mathematical model. The predictions of minimum root mean square error of the 3 kinds of rock composition are： 0.0880,0.0956 and 0.1212.The experimental results show that to determine the content of various mineral composition in natural rock by near-infrared diffuse reflection spectrum is feasible,which vrovides theory basis for the rapid detection of rock components in the future.
出处
《企业技术开发》
2014年第1期42-44,共3页
Technological Development of Enterprise
关键词
近红外漫反射光谱
标准归一化
随机森林
最小均方根误差
岩石矿物
near-infrared diffuse reflection spectrum
standard normalized
random forests
minimum root mean square error
rockand mineral
