基于近红外光谱技术和优化预处理方法的不同品牌燕麦无损鉴别分析

目的 建立红外光谱预处理方法以实现对不同品牌燕麦的快速无损鉴别。方法 通过对不同品牌燕麦样品近红外光谱进行采集, 结合单一以及组合光谱预处理方法消除光谱干扰, 最后利用主成分分析法构建稳健的鉴别模型。结果 光谱预处理有效消除了变动背景以及基线漂移干扰, 最佳预处理方法为二阶导数, 可以实现进口、国产、劣质燕麦之间的完全鉴别。结论 通过近红外光谱预处理方法对原始光谱图进行处理, 可以提取光谱图中有用信息, 消除样品自身不均等因素对光谱的影响, 提高不同品牌燕麦鉴别的准确率。

Non-destructive identification of different brands of oats based on near-infrared spectroscopy and optimized pretreatment methods

Objective To establish a infrared spectrum pretreatment method for the rapid and nondestructive identification of different brands of oats. Methods The near infrared spectra of different brands of oat samples were collected. The spectral interferences were eliminated by single and combined spectral pretreatment methods, and a robust identification model was obtained by principal component analysis method. Results Spectral pretreatment effectively eliminated the variation background and baseline drift interference. The best pretreatment method was the second derivative, which could realize the complete identification among imported, domestic and inferior oats. Conclusion The useful information in the original spectrogram can be extracted by the pretreatment method of near-infrared spectrum, which can eliminate the influence of uneven factors on the spectrum of the sample, and improve the accuracy of identifying different brands of oats.