野生食用菌干品长时间储藏会引起微生物增殖、物理及化学变化，影响其商品品质，为保证其质量安全，亟需建立快速有效的方法，鉴别不同储藏年限野生食用菌。本研究采集5个储藏年限，77个绒柄牛肝菌子实体的紫外（UV）与傅里叶变换红外（FT-IR）光谱，采用卷积平滑（SG）、二阶导数（2-D）、标准正态变量（SNV）等方法对光谱进行预处理，结合偏最小二乘判别分析（PLS-DA），建立UV、FT-IR、低级和中级数据融合模型。结果显示：UV与FT-IR光谱最佳预处理分别为SG+2-D和SG+2-D+SNV；UV、FT-IR、低级和中级数据融合模型，总样品分类错误数分别为10、6、4、3，且中级数据融合的R2cal平均值最接近于1、RMSECV平均值最小，表明中级数据融合分类效果，优于UV、FT-IR和低级数据融合。采用UV与FT-IR中级数据融合策略结合PLS-DA，能够准确鉴别不同储藏年限牛肝菌样品，为野生食用菌品质评价提供一种新思路。

Long term storage of wild edible mushrooms would cause microbial proliferation and physico-chemical changes, affecting the quality. In order to ensure the security and quality, it was essential to establish a quick and efficient method to identify wild edible mushrooms with different storage periods. Ultraviolet (UV) and fourier transform infrared (FT-IR) spectra of 77 fruit bodies of B. tomentipes (5 years of storage) were preprocessed by using Savitzky-Golay (SG) smoothing, second derivative (2-D) and standard normal variate (SNV), and UV, FT-IR, low-level and mid-level data fusion models were established with partial least squares discriminant analysis (PLS-DA). The results showed that the optimal pretreatment of UV and FT-IR spectra were SG+2-D and SG+2-D+SNV, respectively, and the classified individual errors were 10, 6, 4 and 3 in UV, FT-IR, low-level and mid-level data fusion models.Theaverage of R2cal in mid-level data fusion model was closest to 1, and the average of RMSECV was minimum , which indicated that the effects of mid-level data fusion model were better than those of other three models. The UV and FT-IR mid-level data fusion strategy combined with PLS-DA could accurately identify the B. tomentipes with different storage periods, which provided a novel reference for quality evaluation of wild edible mushrooms.

国家自然科学基金项目（31660591、21667031）；云南省教育厅科学研究基金项目（2016ZZX106）；云南省高校食用菌资源开发与利用重点实验室建设项目资助