基于近红外光谱和随机森林方法鉴别蜂蜜真伪

目的建立蜂蜜样品真伪鉴别的近红外光谱快速检测方法,为今后蜂蜜检验工作提供可靠参考依据。方法采用积分球透反射模式采集样品近红外光谱数据,以Savitzky-Golay 1阶微分方法对原始光谱进行预处理,以随机森林方法建立光谱数据与蜂蜜真伪的定性判别模型。结果所建立的判别模型中训练样本判别正确率为100%,测试样本判别正确率为95%。结论近红外透反射光谱技术应用于蜂蜜真伪鉴别的可行性,同其他分析方法相比具有操作简单、速度快、效率高、无污染、费用低、无需复杂前处理等优点。

Detection of honey adulteration by near infrared spectroscopy coupled with random forest method

Objective The combination of near infrared spectroscopy with chemometrics provides an approach to study honey adulteration problem. Methods transflectance mode with integrating sphere was used to collect NIR spectra. Savitzky-Golay first derivative method was employed to preprocess the raw NIR spectra. A Random Forest model was established based on the correlation between the NIR spectra and reference value. Results The prediction accuracy of the proposed method was 100% for training samples and 95% for test samples. Conclusion The proposed method was fast and non-destructive, and provides a novel efficient and environmental friendly approach for the rapid qualitative detection of honey adulteration.