论文部分内容阅读
采用AntarisⅡ傅里叶近红外光谱技术结合模式识别方法研究鉴别天然彩色棉真伪的可行性。原始光谱经多元校正方法(MSC)消除样本的散射和基线漂移等对光谱的影响后,进行系统聚类分析,聚类分析结果显示天然绿色棉和染色绿色棉能得到清晰地分类。通过不同光谱预处理方法,采用簇类的独立软模式法(SIMCA)建立分类模型,结果显示通过对样品光谱中心化+5点卷积平滑+去趋势预处理,采用SIMCA方法建立的分类模型正确率最高,分类正确率为96.77%,类I的模型方差为0.001 3,类Ⅱ的模型方差为0.000 8。研究表明,AntarisⅡ傅里叶近红外光谱技术可用于天然彩色棉真伪的鉴别。
Antaris Ⅱ Fourier Near Infrared Spectroscopy with Pattern Recognition Method to Identify the Authenticity of Natural Colored Cotton is Feasible. The original spectra were analyzed by the system of cluster analysis after eliminating the sample scattering and baseline drift by the multivariate calibration method (MSC). The cluster analysis showed that the natural green cotton and the dyed green cotton can be clearly classified. The classification model was established by different spectral pretreatment methods using cluster-independent soft-mode method (SIMCA). The results showed that the classification model established by SIMCA method was correct by centralizing the sample with +5 point convolutional smoothing and de-trending pretreatment The highest rate was 96.77%, the variance of model I was 0.001 3, and the variance of model Ⅱ was 0.000 8. Studies have shown that Antaris Ⅱ Fourier Near Infrared Spectroscopy technology can be used to identify natural color cotton authenticity.