段焰青, 杨涛, 孔祥勇, 汤丹瑜, 李青青. 样品粒度和光谱分辨率对烟草烟碱NIR预测模型的影响[J]. 云南大学学报(自然科学版), 2006, 28(4): 340-344,349.
引用本文: 段焰青, 杨涛, 孔祥勇, 汤丹瑜, 李青青. 样品粒度和光谱分辨率对烟草烟碱NIR预测模型的影响[J]. 云南大学学报(自然科学版), 2006, 28(4): 340-344,349.
DUAN Yan-qing, YANG Tao, KONG Xiang-yong, TANG Dan-yu, LI Qing-qing. Effects of sample granularity and spectral resolution on tobacco nicotine NIR predoctive model[J]. Journal of Yunnan University: Natural Sciences Edition, 2006, 28(4): 340-344,349.
Citation: DUAN Yan-qing, YANG Tao, KONG Xiang-yong, TANG Dan-yu, LI Qing-qing. Effects of sample granularity and spectral resolution on tobacco nicotine NIR predoctive model[J]. Journal of Yunnan University: Natural Sciences Edition, 2006, 28(4): 340-344,349.

样品粒度和光谱分辨率对烟草烟碱NIR预测模型的影响

Effects of sample granularity and spectral resolution on tobacco nicotine NIR predoctive model

  • 摘要: 以烟草粉末样品的近红外分析为例,研究了在不同样品粒度和不同分辨率下采集的样品光谱对建立的烟碱含量预测模型的影响.结果表明,样品粒度对预测模型有较大影响,细粉末烟叶样品所建的模型在各项评价指标上均较粗粉末样品所建模型好,且实际预测准确性高,因此建议烟叶粉末样品先过0.441 mm以下的筛子后再进行光谱采集.同时,分别以2,4,8,16,32,64cm-1的分辨率扫描的烟叶粉末样品对预测模型的影响也非常明显,高分辨率扫描样品所建的模型在各评价指标上均比低分辨率扫描所建的模型好,预测准确性高,但综合考虑扫描时间和光谱图文件大小等因素,以分辨率为8cm-1扫描的光谱最能满足实际建模的需要.

     

    Abstract: It was studied that the size of sample powders and instrumental resolution on tobacco nicotine NIR predictive model.The results showed that the powder size can affect largely the model.The model based on fine powders can gave more accurate evaluation indexes vs that based on rough powders.Therefore,it is suggested that the tobacco sample for spectra collection should be screened with a 0.441mm or below screen.The scanning sample at resolutions of 2,4,8,16,32 and 64cm-1 showed tat resolution could affect distinctively the model.The model based on higher resolution can give more accurate evaluation indexes vs that based on lower resolution.However,a resolution of 8cm-1 could gve satisfactory result,considering the scanning period and size of spectrogram file.

     

/

返回文章
返回