红外、近红外光谱-簇类的独立软模式方法识别植物调和油脂

用IR,NIR光谱法结合簇类的独立软模式(SIMCA)识别方法对植物油脂进行分类识别,建立了识别二元、三元植物调和油脂的测定方法。应用NIRCal5.2软件的SIMCA技术,分别为所制备的植物调和油脂建立了IR和NIR识别模型,并讨论了光谱处理和数据处理方法来提高模型的分类识别效果。分别以各种植物调和油脂的IR和NIR光谱为变量,随机抽取2/3的样本作训练集,建立了各个调和油的主成分分析(Princi-pal component analysis,PCA)模型;1/3作验证集,对所建模型进行验证识别。用聚类分析-主成分分析(CLU-PCA)方法考察调和油的IR,NIR光谱信息与其纯油的主成分分布。结果显示,在4000～10000cm-1光谱范围内,SIMCA可以对15种二元调和油和2种三元调和油的NIR光谱分别聚类并识别;并对10种二元调和油和2种三元调和油的IR光谱分别聚类并识别。IR以4个波数1099,1119,1746与2855cm-1的吸收值作为分析基础,选择不同的主成分数及数据预处理方法。各种油脂的SIMCA分析的分类精度均为100%,调和油的验证识别准确率100%,最低识别比例为1%,且IR识别灵敏度高于NIR。     

偏最小二乘法在红外光谱识别茶叶中的应用

采用漫反射傅立叶变换红外光谱(FTIR)法结合主成分分析(PCA)、偏最小二乘法(PLS)、簇类的独立软模式(SIMCA)识别法对十三种茶叶进行了分类判别研究。研究结果表明,通过多元散射校正(MSC)对原始光谱进行预处理,可以提高模式识别技术的分类判别效果。在此基础上,选取1 900~900 cm-1波长范围内的茶叶红外光谱建立识别模型,三种方法都得到了满意的分类判别效果。在对检验集中全部130个样本的判别中,PCA仅有两类样本无法判别,SIMCA的识别率和拒绝率都在90%以上,而PLS的识别效果最佳,全部样本都得到了正确的归类。这一研究结果表明傅立叶变换红外光谱法与化学计量学方法相结合可以实现茶叶品种的快速鉴别,这为茶叶的客观评审提供了一种新思路。     

傅里叶变换红外光谱结合模式识别法快速鉴别食用油的真伪

采用傅里叶变换红外光谱(FTIR)结合簇类独立软模式识别技术(SIMCA)建立了真伪食用油的快速鉴别方法. 该方法依据FTIR 的指纹特性, 收集并分析了53 个合格食用油和13 个伪造食用油的FTIR 谱图; 通过对谱图取二阶导数和标准化处理, 主成分分析(PCA)提取特征变量; 采用SIMCA 方法分别随机选取43 个合格食用油和9 个伪食用油样品的FTIR 谱图组成训练集, 构建得到真伪食用油的SIMCA 分类模型. 该模型经过剩余10 个合格食用油和4 个伪食用油的验证, 正确识别率达到了100%. 说明FTIR 结合SIMCA 可能成为快速鉴别食用油真伪的一种新方法.     

基于偏最小二乘回归的类模型方法用于中药牛黄的真伪鉴别

针对独立软模式类簇法(SIMCA)在确定主成分数和决策区间时遇到的困难,提出了一种基于PLSR的类模型方法——PLS类模型方法(PLSCM)。通过把类描述问题转化为常见的PLSR问题,采用成熟的蒙特卡罗交互验证法确定模型的隐变量数和决策区间。采用本方法对不同牛黄样品的近红外光谱数据(波长范围4000～9000 cm-1)进行分析,可成功鉴别牛黄的真伪。本方法的可操作性和鉴别准确率均优于经典的SIMCA方法。对于原始光谱数据,PLSCM的训练和预测准确率均为100%,对于经SNV处理的数据,训练和预测准确率分别为99%和100%。     

“动态”近红外光谱结合深度学习图像识别和迁移学习的模式识别方法研究

该文以山羊绒与山羊绒/羊毛混纺织物以及纯棉与丝光棉织物为研究对象,使用其"动态"光谱,扩大类间的光谱差异信息,通过融合其同步和异步二维相关光谱,用多张动态光谱构造一张能反映细节化学差异信息的"化学图像"。使用GoogLeNet深度神经网络图像识别模型结合迁移学习,建立了一种光谱分类的新方法。收集了234个织物样品,制备水含量分别为0、5.4%、11.2%和16.3%的样本,同时采集样品的漫反射近红外光谱。使用干基样品的多种预处理光谱,利用线性分类方法簇类独立软模式识别(SIMCA)和非线性方法支持向量机(SVM),共建立了16个分类模型。其中,山羊绒与山羊绒/羊毛混纺织物的SIMCA和SVM最优预测正确率分别为63.33%和70.09%,纯棉与丝光棉织物的分别为71.02%和72.51%,均不能实现有效分类。新方法对山羊绒与山羊绒/羊毛混纺织物的预测正确率为92.59%,纯棉与丝光棉织物的为94.74%,获得了有效分类。该文首次将图像分类方法用于光谱分类识别,开辟了一种新的研究途径。针对实际应用能收集到的样品属于小样本,不能满足深度学习需要大数据样本的问题,使用迁移学习方法使深度学习框架适应了光谱分类(小样本),为人工智能领域中先进的识别技术用于解决化学问题提供了一个成功示范。     

光谱结合主成分分析和模糊聚类方法的样品聚类与识别

针对紫外光谱结合化学计量学方法快速测定渣油烃族组成模型适应性问题,对渣油光谱进行主成分分析,以主成分得分作为聚类的特征变量进行模糊聚类,建立了光谱结合主成分分析和模糊聚类方法的样品聚类与识别方法和识别,为光谱结合化学计量分析方法中构正模型的正确选择提供了依据。     

气相色谱结合化学计量学区分大米贮藏时间与产地

香气是衡量大米质量的一个主要因素,对大米的食用品质有重要影响。该文以顶空固相微萃取(SPME)技术为基础,采用气相色谱法分别分析了不同贮藏时间和不同产地大米样本的挥发性成分,通过主成分分析法(PCA)和偏最小二乘判别分析法(PLS-DA)对大米样本进行分类和判别分析。PCA及PLS投影图显示不同储藏时间的大米明显聚为4类,通过留一交叉验证法(LOO)计算PLS预报的准确率为96%,相对标准误差为8.2%。同时,PCA投影图中可将4个不同产地的大米样本进行区分,分类效果显著;所建PLSDA模型可靠,不同产地大米样本均能被准确识别,正确率为100%。以顶空固相微萃取/气相色谱检测大米中挥发性成分,利用主成分分析法和偏最小二乘判别分析法鉴别大米新鲜程度和产地具有可行性。     

基于红外光谱多元分析的记号笔种类鉴别研究

借助衰减全反射红外光谱结合K近邻算法和层次聚类,建立记号笔墨水种类鉴别的分类模型,为记号笔的种类鉴别提供有效的新方法。研究基于记号笔墨水的外光谱矩阵数据,通过建立主成分和判别分析分析模型和K近邻算法分类模型,实现对模型性能的比较和对模型分类结果的验证。实验结果表明,以水性和油性作为分类标准,模型对样本的区分能力好,其正确率为100%。借助红外谱图进一步分析水性油性样本时,其最强峰与其对应溶剂相符合。借助K近邻算法进行验证性分析,按重要性加权特征给不同的样本施加不同的权重,运用训练样本即为测试样本的方法交互验证,选取K值为1,训练集∶保持集=3∶1,建立分类模型,模型总分类准确率达100%,区分效果良好,不同品牌的油性、水性样本能被聚类为一组。综上,衰减全反射红外光谱结合K近邻算法和层次聚类可作为记号笔墨水种类鉴别的一种快速准确的分析方法。     

红外光谱结合主成分分析鉴别道地山药

用红外指纹图谱和主成分分析法，对来自8省的74个山药样本的道地与非道地性进行了聚类分析，以主成分2和3作二维线性投影图，均具有较好的聚类和鉴别作用。利用已建模型预测了11个样本的道地性，正确率达90.9%。结果表明：所建模型合理，具有明显的分类和鉴别作用，具有一定的可行性和可靠性，为山药道地性鉴别提供了一种新的方法。     

近红外光谱法定性检测菜籽油中多效唑残留

采用近红外光谱技术结合化学计量学方法对菜籽油中多效唑残留进行定性检测。在4000~10000 cm-1光谱范围内采集126个菜籽油样本的近红外透射光谱。对原始光谱进行初步分析后,分别采用线性判别分析(LDA)、簇类独立软模式法(SIMCA)和最小二乘支持向量机(LSSVM)三种不同方法建立菜籽油中多效唑残留的定性检测模型,并对不同多效唑残留的菜籽油样本的分类正确率进行分析。研究结果表明,LDA,SIMCA及LSSVM 3种方法建立的检测模型均具有较高的判别能力,其校正集和预测集的正确率分别为93.33%,91.11%,95.56%和86.11%,88.89%,83.33%。此外,高多效唑残留样本的分类正确率大致趋于100%,而低多效唑残留样本的分类正确率则有一定波动。由此可知,利用近红外光谱技术可对菜籽油中多效唑残留进行快速、无损的定性检测。     

Quality assessment of Radix Codonopsis by quantitative nuclear magnetic resonance

Radix Codonopsis (Dangshen) is a famous traditional Chinese medicine and has long been used for replenishing energy deficiency, strengthening the immune system, lowering blood pressure and improving appetite in China, Japan and Korea. A highly specific quantification method using ¹H NMR has been developed for the simultaneous determination of novel quaternary ammonium alkaloids codotubulosine A and B, adenosine and 5-(hydroxymethyl)furfural in Radix Codonopsis materials Codonopsis pilosula, C. pilosula var. modesta, C. tangshen, C. tubulosa, C. subglobosa, C. clematidea, C. lanceolota and Campanumoea javanica collected from different regions of China and Taiwan. A solid-phase extraction with C-18 cartridge followed by elution with water can easily remove sugars the major components that may affect the determination of target constituents. In the ¹H NMR spectrum, the signals of N-CH₃ of codotubulosine A (δ 2.75) and B (δ 2.83), H-8 of adenosine (δ 8.15), and CHO signal of 5-(hydroxymethyl)furfural (δ 9.49) are well separated from other signals in [²H₄]methanol. The quantity of the compounds was calculated by the relative ratio of the integral values of the target peaks of each compound to the known amount of internal standard pyrazine. The described NMR method is found to be relatively simple, specific, precise and accurate for the quality control of Radix Codonopsis herbs and no reference compounds are required for calibration curves, in comparison to conventional HPLC methods, for instance.     

Analysis of 14 terpenoids and sterols and variety discrimination of Codonopsis Radix using ultra-high-performance liquid chromatography-quadrupole-time-of-flight mass spectrometry

Recently, we confirmed that the 95% ethanol-extracted fraction of Codonopsis Radix, which contains several triterpenoids and sterols, possesses pharmacological activities. However, due to the low content and diverse types of triterpenoids and sterols, their similar structures, lack of ultraviolet absorption, and difficulty in obtaining controls, few studies have so far assessed their contents in Codonopsis Radix. We accordingly constructed an ultra-high-performance liquid chromatography-quadrupole-time-of-flight mass spectrometry technique for the simultaneous quantitative determination of 14 terpenoids and sterols. Separation was performed on the Waters Acquity UPLC HSS T3 C₁₈ column (100 × 2.1 mm, 1.8 μm) with 0.1% formic acid (A) and 0.1% formic acid in methanol (B) as mobile phase under gradient elution. The determination coefficients for each of the matrix calibration curves were ≥0.9925. The average recovery ranged from 81.25% to 118.05%, with relative standard deviations of <4%. The contents of 14 components in 23 batches were quantified and further analyzed through chemometrics. Linear discriminant analysis can distinguish sample varieties. The quantitative analysis method can accurately determine the contents of 14 components and thereby provide the chemical basis for the quality control of Codonopsis Radix. It also could be a valuable approach for the classification of different Codonopsis Radix varieties.     

基于近红外光谱分析技术的食品包装塑料的定性分析

该文基于近红外漫反射光谱分析技术对食品包装材料聚乙烯、聚丙烯进行定性判别试验研究,选取不同波段范围、采用不同光谱预处理方法,使用主成分分析法(Principal component analysis,PCA)结合SIMCA、贝叶斯判别、K-近邻3种模式识别方法建立定性预测模型,并根据正确识别率比较了各模型预测性能。结果表明：使用SIMCA方法、贝叶斯判别、K-近邻3种方法建立的定性校正模型均在1 050~1 550 nm波长范围内效果较好;采用矢量归一化、标准正态变量变换、中心化、滑动均值滤波、多项式平滑滤波、一阶微分6种光谱预处理方法和上述3种模式识别方法对塑料样品近红外光谱进行了数据处理,其中在1 050~1 550 nm范围内,主成分因子数为3,采用原始光谱建立的K-近邻定性校正模型较优,对样品校正集和预测集的正确识别率均为100%。可为食品包装材料聚乙烯、聚丙烯的快速鉴别研究提供参考。     

Evaluation of surfactant‐assisted pressurized hot water extraction for marker compounds in Radix Codonopsis pilosula using liquid chromatography and liquid chromatography/electrospray ionization mass spectrometry

In the move towards the elimination of organic solvents in the extraction process in botanicals, a new method combining surfactant and pressurized hot water extraction (PWHE) with an applied temperature below the boiling point and lower pressure from 10 to 20 bar was developed for the analysis of marker compounds that are reasonably hydrophobic such as tetradeca‐4E,12E‐diene‐8,10‐diyne‐1,6,7‐triol and tetradeca‐4E,12E‐diene‐8,10‐diyne‐1,6,7‐triol‐O‐β‐D‐glucoside in Radix Codonopsis pilosula (DangShen). Because reference substances for the proposed botanicals were not available, a method was developed to isolate the marker compounds in Radix Codonopsis pilosula. Other than surfactant‐assisted PHWE, the marker compounds present in Radix Codonopsis pilosula were extracted using pressurized liquid extraction (PLE) with methanol and PHWE with a mixture of water/ethanol (80:20). The extracts were analyzed using liquid chromatography and liquid chromatography/electrospray ionization mass spectrometry. With surfactant‐assisted PHWE, the effects of different added surfactants such as sodium dodecyl sulfate and Triton X‐100 was studied. Surfactant assisted PHWE with Triton X‐100 proved to be at least equivalent or better compared to Soxhlet extraction in terms of quantitative analysis of marker compounds in Radix Codonopsis pilosula. The method precision was less than 8% (RSD, n = 6). The presence of surfactants in PHWE was found to enhance the solubility of target compounds naturally occurring in medicinal plants.     

Characterization of chemical composition variations in raw and processed Codonopsis Radix by integrating metabolomics and glycomics based on multiple chromatography-mass spectrometry technology

Codonopsis Radix, a popular food homology medicine, is widely used in clinical traditional Chinese medicine and food supplement, raw products and three types of processed products are the main forms of decoction pieces in China. However, there is no scientific basis for comprehensive chemical characterization of raw and three types of processed products. Herein, we investigated qualitatively and quantificationally secondary and primary metabolites in raw Codonopsis Radix and three types of processed products by metabolomics and glycomics employing multiple chromatography-mass spectrometry technology combined with chemometric analysis further to look for differential compounds and propose the processing-induced chemical mechanisms. The results indicated that Codonopsis Radix became dark-colored and the smell of burnt incense odor was observed after processing. The principal component analysis demonstrated that secondary metabolome and glycome were significantly altered between raw and processed products, and 36 differential secondary metabolites and 11 differential primary metabolites were finally screened through orthogonal partial least-squares-discriminant analysis. The main types of compounds are alkaloids, terpenoids, glycosides, amino acids, monosaccharides, oligosaccharides, and furfural derivatives. Meanwhile, Chemical mechanisms could be involved, including oxidation, glycosidic hydrolysis, esterification, dehydration, and Maillard reaction. This work supplies a chemical basis for the application of various types of Codonopsis Radix decoction pieces.     

葛根药材指纹图谱分析方法研究

以安徽产葛根药材为分析对象,选择Agilent ZORBAX SB-C18柱(4.6 mm×250 mm,5μm),甲醇-水梯度洗脱,流量为0.9 mL.min-1,检测波长为250 nm,柱温25℃,建立了葛根药材HPLC指纹图谱。结果表明,10批供试品的保留时间在60 min之内,共有18个典型的共有色谱峰。采用HPLC法获得的葛根药材指纹图谱稳定、可靠、重复性好,能较全面地反映葛根药材的内在质量,可用于葛根药材的质量控制。     

Simultaneous chemical fingerprinting and quantitative analysis of crude and processed Radix Scrophulariae from different locations in China by HPLC

A validated liquid chromatography method was first developed to evaluate the quality of crude and processed Radix Scrophulariae extracts through establishing chromatographic fingerprint and simultaneous determination of five bioactive compounds, namely 5-hydroxymethylfurfural (5-HMF), acteoside, angroside C, harpagoside and cinnamic acid. The chromatographic were separated on an Agilent Zorbax Extend C(18) column (250 mm × 4.6 mm, 5 μm) and detected by diode array detector (DAD). Mobile phase was composed of (A) aqueous phosphoric acid (0.03%, v/v) and (B) acetonitrile using a gradient elution. Analytes were performed at 30 °C with a flow rate of 1.0 mL/min and UV detection at 280 nm. All calibration curves showed good linear regression (r(2) ≥0.9996) within the tested ranges, and the recovery of the method was in the range of 98.12-103.38%, with RSD values ranging from 0.6 to 2.8%. In addition, the contents of those five bioactive compounds in crude and processed Radix Scrophulariae prepared by different locations of China were determined to establish the effectiveness of the method. The results demonstrate that the developed method is accurate and reproducible and could be readily utilized as a suitable quality control method for the quantification of Radix Scrophulariae.     

Comparison of reversed‐phase liquid chromatography and hydrophilic interaction chromatography for the fingerprint analysis of Radix isatidis

Radix isatidis is a famous anti‐influenza virus herbal medicine traditionally taken as a water decoction. However, the chemical fingerprint analysis of Radix isatidis is dominantly based on RPLC, from which it is difficult to obtain fingerprint information of hydrophilic compounds. Here, we developed the separation of Radix isatidis by RPLC and hydrophilic interaction chromatography, comparing the traditional RPLC fingerprint with the hydrophilic interaction chromatography fingerprint. Besides, an anti‐viral assay of Radix isatidis was conducted to evaluate its efficacy. The fingerprint–efficacy relationships between the fingerprints and the anti‐viral activity were further investigated with principal component regression analysis. The results showed that the anti‐viral activity correlated better with the hydrophilic interaction chromatography fingerprint than with the RPLC fingerprint. This study indicates that hydrophilic interaction chromatography could not only be a complementary method to increase the fingerprint coverage of conventional RPLC fingerprint, but also can better represent the efficacy and quality of Radix isatidis.     

丹参水溶性成分的电喷雾质谱行为及其特征图谱的初步研究

应用电喷雾-质谱(ESI-MS)技术研究了原儿茶醛和丹参素的ESI-MS规律,结合ESI-MS技术,对9种丹参水溶性成分的质谱行为进行了研究,并对其中5种成分进行了定性分析。同时利用选择性离子检测质谱技术,初步建立了丹参水溶性成分SIM-MS特征图谱,该图谱简单、易于解析,重现性与特征性较好,可以用于丹参药材及复方丹参的快速指纹鉴别。