随着人民生活水平的提高, 中草药的保健功能越来越得到大家的重视, 铁皮石斛是我国名贵中药材, 素有“救命仙草”之称。 尝试用叶绿素、 糖度和pH值作为铁皮石斛的品质指标, 选择安徽霍山、 浙江雁荡山和云南3个不同产地的铁皮石斛作为研究对象, 提取不同铁皮石斛的光谱数据和理化参数, 然后进行各品质指标的反演, 最终建立品质指标与光谱之间的相关模型, 并对模型的有效性进行分析。 实验中的研究对象是铁皮石斛的茎, 用ASD光谱仪得到三种铁皮石斛茎的光谱数据, 再把同一批样本研磨碎, 放入离心管中加入甲醇溶液密封, 并用锡箔纸包装处理制成相应的溶液, 采用分光光度计、 糖度计和pH计测量叶绿素含量、 糖度和pH值, 每个样本分别选取离心试管中的上层, 中层, 下层溶液检测糖度, 测量3次并取平均值, 以此结果作为对照组。 原始光谱数据通过小波变换去除噪声和降维, 将得到的能量系数(包括波段与尺度), 与铁皮石斛对照组的理化参数进行相关性分析, 选取决定系数中较高的能量系数作为小波特征, 用最小化二乘法对小波特征拟合。 用全部实验样本作为测试集, 随机选取70%作为验证集, 对于浙江雁荡山、 安徽霍山和云南三种不同品种铁皮石斛: 叶绿素含量反演模型的决定系数( R²)分别为0.819, 0.820和0.865, 均方根误差(RMSE)分别为为0.035, 0.013和0.017; 糖度反演模型的决定系数( R²)分别为0.756, 0.764和0.823, 均方根误差(RMSE)分别为0.025, 0.030和0.0368; pH值的反演模型决定系数( R²)分别为0.819, 0.820和0.865, 均方根误差(RMSE)分别为0.0345, 0.013和0.017。 从中可以发现, 三种不同铁皮石斛的品质反演模型和决定系数( R²)均大于0.80, 均方根误差(RMSE)小于0.10。 实验证明了铁皮石斛中叶绿素、 糖度和pH值的光谱特性对其进行品质评估的可行性。

With the improvement of people’s living standards, people pay more and more attention to the health care function of Chinese herbal medicine. Dendrobium officinale Kimura et Migo (Dendrobium officinale) is a rare Chinese herbal medicine, known as “life-saving fairy grass”. In this study, we tried to evaluate the quality of Dendrobium officinale by sugar content, pH value and other related physical and chemical characteristics. We selected three different habitats of Dendrobium officinale from Huoshan in Anhui, Yandang Mountain in Zhejiang and Yunnan Province as research objects, extracted the spectral data and physical and chemical parameters of different Dendrobium officinale, and then carried out the inversion of quality indicators. Finally, the correlation model between quality and spectrum was established. In the experiment, the leaves, roots and flowers of Dendrobium were removed first, and then the stems to be studied. The spectral data of three Dendrobium officinale with different quality grades were obtained by the ASD specter. The same samples was ground, put into a centrifugal tube, sealed with methanol solution, and packaged with tin foil paper to make corresponding solutions. The chlorophyll content, sugar content and pH value were measured by spectrophotometer, sugar meter and pH meter. The upper layer, middle layer and lower part of the centrifuge tube were selected for each sample. Each sample was measured 3 times, and the average value was taken as the control sample. The original spectral data were denoised and dimensionally reduced by wavelet transform. The correlation between the energy coefficients (including band and scale) and the physicochemical parameters of the Dendrobium officinale control group was analyzed. The higher energy coefficient in the determination coefficient was selected as the wavelet feature, and the wavelet feature was fitted by the least square method. Using all experimental samples as test set and 70% as verification set, the determination coefficient ( R²) of chlorophyll content inversion model was 0.819, 0.820 and 0.865, the root mean square error (RMSE) were 0.035, 0.013 and 0.017, respectively; the determination coefficient ( R²) of sugar content inversion model was 0.756, 0.764 and 0.823, respectively. The results showed that the root means square error (RMSE) was 0.025, 0.030 and 0.036 8; the determination coefficient (R²) of the inversion model for pH value was 0.819, 0.820 and 0.865, and the root mean square error (RMSE) was 0.034 5, 0.013 and 0.017, respectively. It can be found that the quality inversion model and determination coefficient ( R²) of three kinds of Dendrobium officinale are all greater than 0.80, and the root means square error (RMSE) is less than 0.10. This study proved that the spectral characteristics of chlorophyll, sugar content and pH value in Dendrobium officinale were feasible for quality evaluation.