数据融合快速鉴别9种野生食用牛肝菌

食用菌种类与产地品质存在差异,加上因牟利产生的混杂销售现象,严重制约了高原特色农产品来源鉴别、种质资源评价和深入挖掘利用。本试验融合牛肝菌光谱信息建立支持向量机（SVM）模型,寻找最佳的样品种类鉴别方法。结果显示：（1）元素标准曲线R ²>0.999,RSD<5.0%,标准物回收率94%-106%,测定方法可靠;（2）样品含Ca、Na等人体必需元素,但Cd含量超标;（3）脂肪酸、蛋白质等化合物和Ni、Co等矿质元素对种类鉴别贡献最大;（4）中级数据融合优于低级数据融合,优于单一光谱数据模型。数据融合结合化学计量学可实现样品种类快速准确鉴别,对食用菌市场监督、种质资源评价及挖掘利用具有理论参考意义。

Rapid quality identification of nine wild edible Boletus mushrooms by using data fusion

The prediction model of support vector machine (SVM) was established based on spectra and data fusion to find reliable approach for quality identification of marketing edible Boletus in Yunnan Province. The results indicated that element content calibration curve was R ²>0.999 and RSD<5.0%, and the standard recovery rate was 94%-106%. The marketing samples contain necessary elements of human body such as Ca, Na, etc., while Cd content exceeds the provided standard. Fatty acids, proteins, Ni, Co, etc. were the main substances for quality identification. Among all the models, mid-level data fusion was superior to low-level data fusion and single spectral data model. The rapid and accurate quality identification was achieved by means of data fusion combined with chemometrics, providing a theoretical reference for market supervision, germplasm resources evaluation and exploiting utilization potentiality of Boletus mushrooms.