基于粒子群算法的波长选择方法用于苹果酸度的近红外光谱分析

采用便携式近红外光谱分析仪,对苹果样品进行扫描获得光谱数据,运用偏最小二乘法结合基于粒子群算法的波长选择方法对苹果试验数据进行多元统计分析,建立数学模型,利用该模型对苹果酸度进行了预测。对于基于粒子群算法和全谱偏最小二乘方法,校正集样品的酸度预测值和实测值之间的相关系数分别为0.9880和0.9553,校正均方根误差分别为0.0197和0.0388;预测集样品的酸度预测值和实测值之间的相关系数分别为0.9833和0.9596,预测均方根误差分别为0.0193和0.0304。与全谱偏最小二乘法相比,基于粒子群算法的偏最小二乘法,不仅较大地减少波长变量而降低计算量,而且也较大地提高了模型性能而增强了模型预测的准确性。该方法可建立较好的定量分析模型,能广泛应用于现场或野外苹果酸度的快速分析。     

废旧聚酯/棉混纺织物的在线近红外定量分析与自动分选

在前期探究的最佳测试条件下,利用自主研制的“纤维制品主体组分高效识别与分选装置”对废旧聚酯/棉混纺织物样品进行在线原始近红外光谱采集。基于在线原始谱图,探讨出最佳光谱预处理方法为S-G平滑+最大最小归一化(MMN)+S-G导数,并利用偏最小二乘法建立了废旧聚酯/棉混纺织物的在线近红外定量分析模型,模型的交互验证均方根误差(RMSECV)为1.47,校正相关系数(RC)、验证相关系数(RV)值均不小于0.99,校正相对预测偏差(RPDC)为18.17,验证相对预测偏差(RPDV)为13.13,交互验证相对预测偏差(RPDCV)为11.76。为验证模型的可靠性,选取30个外部样本进行在线验证,验证结果的线性方程为y=(1.00±0.01)x-(0.88±0.56),预测准确率为93.3%。将模型导入分选装置的“纺织品在线主控程序”后,对设备设定不同聚酯含量织物的分选类别,即可对废旧聚酯/棉混纺织物样本进行含量预测,并通过装置的吹分分选系统将样品自动吹扫到相应的收集框中。每个样品预测并分选的时间小于2 s,机械自动分选结果无误。利用所建模型和分选装置可对废旧聚酯/棉混纺织物进行在线高效测定与自动分选。     

基于近红外光谱技术的亚麻纤维化学成分含量快速测定

在推进亚麻纤维的纺纱及其产业化生产过程中,快速、准确的定量分析纤维的化学成分是重要趋势。该研究利用近红外光谱技术分析亚麻纤维化学成分,以化学分析法测定值为对照,采用偏最小二乘法(PLS)建立亚麻纤维化学成分的近红外模型,从而实现了其化学成分的高效、快速定量分析。结果表明,建立的亚麻纤维纤维素、半纤维素、木质素和果胶近红外模型的校正相关系数(R_C)与验证相关系数(R_(CV))均在0.9以上,校正均方根误差(RMSEC)小于预测均方根误差(RMSEP)且均小于1。外部验证和双尾t检验表明模型预测结果较为准确,预测值与化学分析法得到的实测值无显著性差异,故该模型可用于相关化学成分含量的快速预测。     

藏药五脉绿绒蒿提取过程的在线近红外光谱质量控制研究北大核心CSCD

利用近红外光谱技术和自建的在线检测系统,实现了藏药五脉绿绒蒿提取过程中总黄酮含量的在线近红外光谱监测和提取终点的判定。以403个样品为建模集,分别获得了主成分回归(PCR)、偏最小二乘(PLS)、决策树(DT)、随机森林(RF)算法下的最佳光谱预处理方法和建模区间,以残差预测偏差(RPD)值为指标选择最佳建模方法。以62个样品为外部验证集,考察模型应用于总黄酮含量实时监测的可行性。此外,还探讨了利用模型预测值进行相对浓度变化率(RCCR)分析直接判定提取终点的可行性,并比较了标准偏差绝对距离法(ADSD)和移动窗口标准偏差法(MBSD)对提取终点判定的适用性。结果表明,在预处理方法为Constant+一阶导数+SG平滑、建模区间5300~9000 cm^(-1)条件下所建的总黄酮含量的PLS模型效果最好,其校正集和验证集的误差均方根均小于0.14、相关系数均大于0.97,RPD值为4.68。所建PLS模型对未知样品的平均预测率为79%,实际值与预测值的相关系数大于0.98,表明模型有较好的预测效果。外部验证集中RCCR法判定的预测提取终点和ADSD法判定的提取终点均与实际提取终点一致。所建模型性能较好,通过对未知样品进行准确快速的定量分析,实现了五脉绿绒蒿提取过程中总黄酮含量的实时监测,同时,以RCCR和ADSD作为提取终点的判定方法较为准确,可为藏药材提取过程在线近红外光谱分析技术的研究提供有益借鉴。     

近红外光谱法快速检测烟草中部分香气物的应用研究

应用傅立叶变换近红外漫反射光谱仪,从500个样品中按高、中、低含量挑选出具代表性的150～172个烟草样品建立了近红外光谱与烟草中的苹果酸、柠檬酸和石油醚提取物成分含量间的数学模型,用建立的模型对36个样品进行预测,结果表明,各成分近红外预测值与实测值之间的平均偏差:苹果酸为0.090,柠檬酸为0.040,石油醚提取物为0.124;且近红外预测值与化学法不存在显著性差异,近红外光谱分析技术可初步用于烟草部分香气成分的快速定量分析.     

近红外光谱技术对烟草pH值的快速测定

提出了一种基于近红外漫反射光谱技术快速测定烟草pH值的方法.采集不同烟草粉末样品的近红外漫反射光谱,并对其原始光谱数据进行一阶微分、二阶微分及平滑等预处理,用偏最小二乘法(PLS)方法建立pH值预测模型(建模样品572个).从光谱主成分分布和pH值分布方面考察了81个验证集样品,结果表明验证集样品分布范围较大,分布较合理.利用主仪器模型对验证集样品进行预测,结果表明主仪器一阶微分模型和二阶微分模型对验证集样品的pH值预测与实际测量值的平均绝对偏差分别为0.057、0.065,t检验表明预测值和实测值之间没有显著性差异,达到了较好的结果.考察了主仪器pH值一阶微分、二阶微分模型在同一型号不同仪器间的直接转移效果,一阶微分模型转移给了子仪器A ～F,二阶微分模型转移给了子仪器G,7台子仪器pH值预测的平均绝对偏差为0.049 ～0.070,且都通过了F检验.实验表明,该主仪器模型能够快速预测烟叶中的pH值,并成功转移到同类仪器上进行检测.     

鲜辣椒中糖份和维生素C含量的近红外光谱非破坏性测定

用近红外光谱法非破坏测定鲜辣椒中可溶性糖和维生素C含量，可溶性糖含量的化学值与近红外预测值之间的相关系数为0．9024，校正集标准差(SEP)为1．23％，RSD9．3％；VC含量的化学值与近红外预测之间的相关系数为0．9122，校正集标准差(SEP)为24．17，RSD为9．5％，辣椒鲜果中可溶性糖和VC含量与近红外光谱有显著的相关关系。     

基于BP-神经网络的航空煤油总酸值近红外光谱快速检测

针对航空煤油中总酸值量较小,在近红外定量分析时有用信息易被干扰的问题,采用误差反向传播神经网络(BP-ANN)建立航空煤油总酸值近红外光谱分析模型.根据模型校正集预测偏差最小原则,确定了隐含层神经元个数、学习效率等参数.用建立的网络模型预测了验证集样品总酸值,预测的相关系数R2为0.9778,预测标准偏差(RMSEP)...     

川贝母中浙贝母掺入量的快速检测方法研究

外观相似、化学成分相近的中药材,很难进行有效的质量控制.本文应用近红外漫反射光谱技术和化学计量学手段,建立了快速检测川贝母中浙贝母掺入量的新方法.以含不同川贝母和浙贝母比例的41个样本建立校正集,通过偏最小二乘法得到最佳校正模型.对6个预测集样本进行预测分析,得到近红外光谱预测值与真值的相关系数r=0.9997,样品回收率为97.96%～100.86%,RSD为0.8l%.本方法具有快速方便、结果准确的特点,可以应用于中药材的品质分析和质量控制中.     

木薯品质分析的近红外光谱模型建立及其应用研究

应用近红外光谱法(NIRS)建立木薯中淀粉、水分定量分析的近红外光谱数学模型,探讨了修正偏最小二乘法(MPLS)、偏最小二乘法(PLS)以及主成分回归法(PCR)等优化处理对定标模型的影响,确定了修正偏最小二乘法(MPLS)是建立模型最适合的数学方法。并对模型预测结果的准确性进行了评价。结果表明:验证集样品的化学值和近红外预测值拟合存在较好的线性关系,相关性显著。淀粉模型预测标准偏差(Sep)为0.850,系统偏差(Bias)为-0.095,相关系数(r)为0.971。水分模型预测标准偏差(Sep)为0.075,系统偏差(Bias)为0.007,相关系数(r)为0.980。淀粉、水分定量分析的NIRS数学模型具有较高的预测准确性,可应用于木薯批量收购中的品质等分析。     

Rapid quality assessment of Gentianae Macrophyllae Radix based on near-infrared spectroscopy and capillary electrophoresis

The aim of this study was to establish a rapid quality assessment method for Gentianae Macrophyllae Radix (RGM) using near-infrared (NIR) spectra combined with chemometric analysis. The NIR spectra were acquired using an integrating sphere diffuse reflectance module, using air as the reference. Capillary electrophoresis (CE) analyses were performed on a model P/ACE MDQ Plus system. Partial least squares-discriminant analysis qualitative model was developed to distinguish different species of RGM samples, and the prediction accuracy for all samples was 91%. The CE response values at each retention time were predicted by building a partial least squares regression (PLSR) calibration model with the CE data set as the Y matrix and the NIR spectra data set as the X matrix. The converted CE fingerprints basically match the real ones, and the six main peaks can be accurately predicted. Transforming NIR spectra fingerprints into the form of CE fingerprints increases its interpretability and more intuitively demonstrates the components that cause diversity among samples of different species and origins. Loganic acid, gentiopicroside, and roburic acid were considered quality indicators of RGM and calibration models were built using PLSR algorithm. The developed models gave root mean square error of prediction of 0.2592% for loganic acid, 0.5341% for gentiopicroside, and 0.0846% for roburic acid. The overall results demonstrate that the rapid quality assessment system can be used for quality control of RGM.     

Rapid Determination of the Main Compounds in Crude and Processed Atractylodes macrocephala Using Fourier Transform Infrared Spectroscopy with Attenuated Total Reflectance

Infrared spectroscopy with attenuated total reflectance was investigated for the rapid determination of atractylenolide I and atractylenolide III contained in crude and processed Atractylodes macrocephala. High performance liquid chromatography was used as a reference method for the analysis, and a multivariate calibration model based on a partial least squares algorithm was developed to correlate the spectra and values determined by the reference method. The results suggested that the relative deviation of the predicted contents of atractylenolide I and atractylenolide III in the samples were less than 5.0%, which proved the stability and reliability of the models used. The proposed method is fast and nondestructive, and provides a novel, efficient, and environmentally-friendly approach for the rapid determination of the active components in crude and processed traditional Chinese herbal medicines.     

近红外光谱法测定黄芩提取物中黄芩苷含量

近红外光谱技术(NIR)是近年来快速发展的一种新型光谱分析技术,具有快速、高效、无污染、非破坏性以及实时分析等优点~([1]),已在农业、烟草、石油化工、医药等多领域得到广泛应用.尤其在药物分析方面,体现出近红外光谱分析的巨大潜力~([2]).     

Near infrared reflectance spectroscopy as a tool for the control of sheep leather defatting

The fat content is one of the variables to be controlled by the tanning industry with a view to obtaining leather for various commercial purposes. Ensuring the production of quality leather products frequently entails using some defatting treatment, particularly when the raw skin is rich in natural fat. The official method for determining fat in leather, IUC 4, is rather slow; also, it uses polluting reagents and involves powdering samples for Soxhlet extraction with low-polarity solvents. The combination of NIR diffuse reflectance spectroscopy as implemented with a fibre-optic probe and multivariate calibration is probably the best choice for the direct determination of fat in leather and the monitoring of leather defatting.In this work, a method for the determination of fat in leather and the control of the defatting process in an expeditious manner and with no sample treatment was developed. Defatting tests were conducted on leather specimens from lambs of various breeds and origins in order to span as wide as possible a range of variability in their properties and natural fat content. The NIR spectra used to construct the calibration matrices were recorded directly on the leather samples prior to and after defatting. Fat contents were determined by partial least-squares regression (PLSR), using the values obtained with the official method as references. Notwithstanding the complex nature of leather, the calibration models used provided good external predictions: the largest overall relative error, obtained by using a single calibration matrix for natural and defatted specimens, was 10%. The proposed method is therefore an advantageous alternative to the official method.     

阴离子交换树脂结合近红外漫反射光谱测定碳酸饮料中日落黄

采用强碱性阴离子交换树脂富集饮料中的合成食用色素日落黄,用近红外漫反射光谱技术直接测定富集有色素的树脂.将34个模拟样品建模,日落黄浓度范围为0.05～1.2g/L.以柠檬黄和亮丽春红5R为干扰,经偏最小二乘回归建模,得到决定系数为0.9883,标准偏差为0.0187的稳健模型.定量预测3种不同市售饮料中的日落黄,回收...     

Enhancing calibration models for non-invasive near-infrared spectroscopical blood glucose determination

Partial least-squares regression (PLS) and radial basis function (RBF) networks are used to compute calibration models for non-invasive blood glucose determination by NIR diffuse reflectance spectroscopy. A model computation shows that even extremely small deviations of the spectra induce increased prediction errors. Since the spectral contribution of blood glucose is much smaller than deviations resulting from the non-invasive measuring process a method based on Pearson’s correlation coefficient can be used for evaluating the quality of the recorded spectra during the prediction step. Another method is based on the leverage values from the hat matrix of the RBF network. Both methods lead to a significant decrease in prediction error.     

Comparison between different data pre-treatment methods in the analysis of forage samples using near-infrared diffuse reflectance spectroscopy and partial least-squares multivariate calibration method

Moisture and protein content of alfalfa samples from Catalonia (Spain) have been analyzed by near-infrared (NIR) diffuse reflectance spectroscopy and multivariate calibration methods. In order to remove systematic variation in experimental data, such as base-line and multiplicative scatter effects, the evaluation of different data pre-processing methods is performed. Different figures of merit are used for quality assessment and comparison of these pre-treatment methods.     

总磷测定仪校准方法

建立总磷测定仪的校准方法。介绍了总磷测定仪的工作原理。总磷测定仪的校准包括温度示值误差和温场均匀性、消解时间示值误差、仪器示值误差、测量重复性、示值稳定性5个项目。利用实验数据确定了总磷测定仪的技术指标:温度示值误差不超过±2℃,温场均匀性不大于3℃,消解时间示值误差为不超过±2%,仪器示值误差为不超过±10%,仪器重复性为不大于5%,示值稳定性为不超过±8%FS。该方法可用于总磷测定仪的校准,保证测量数据的准确可靠。     

银镜增强的近红外漫反射光谱法测定血清尿素

针对近红外漫反射光谱(NIRDRS)技术灵敏度低或检出限高的缺点,采用银镜作为吸附基底以改善其灵敏度.银镜的强反射能力不仅能够降低光谱的背景干扰,还能增强光谱的响应信号.研究了NIRDRS技术结合银镜基质用于快速定量分析血清尿素含量的可行性.直接采集富集了血清的银镜基质的NIRDRS光谱,结合光谱预处理和变量选择方法,采用偏最小二乘回归建立了定量校正模型并进行快速预测.结果表明,采用银镜基质结合NIRDRS技术可以准确地测定含量为2.8~26.1 mmol/L的血清尿素,预测值与参考值的相关系数(R2p)为0.9823,样品回收率为86.0%~117.0%,且预测得到的最大误差值低至1.45 mmol/L.     

Rapid Simultaneous Determination of Andrographolides in Andrographis paniculata by Near-Infrared Spectroscopy

The potential of near-infrared spectroscopy (NIRS) for the quality control of traditional Chinese medicine has been evaluated. Seven quantitative parameters, andrographolide, deoxyandrographolide, dehydroandrographolide, neoandrographolide, moisture, ash content, and alcohol-soluble extract of Andrographis paniculata, were evaluated by NIRS. The reference values of andrographolides were determined by high-performance liquid chromatography, and the others were obtained using the standard methods of the 2015 Chinese Pharmacopoeia. The predicted values were determined by a quantitative model using NIRS based on partial least square regression. Different spectral preprocessing methods, spectral ranges, and optimum number of factors were selected to optimize the models. All models were estimated by the combination of various parameters, including the correlation coefficient of calibration for andrographolide, deoxyandrographolide, dehydroandrographolide, neoandrographolide, moisture, ash content, alcohol-soluble extract (values of 0.980, 0.984, 0.989, 0.983, 0.987, 0.988, 0.979, respectively), root mean square error of calibration (values of 0.156, 0.038, 0.050, 0.029, 0.604, 0.431, 0.135, respectively), root mean square error of prediction (values of 0.169, 0.041, 0.050, 0.033, 0.280, 0.493, 0.140, respectively), root mean square error of cross-validation (values of 0.626, 0.114, 0.158, 0.046, 1.145, 0.774, 0.508, respectively), and ratio of standard deviation to standard error of prediction (values of 4.583, 4.690, 4.796, 4.899, 4.899, 4.690, 5.099, respectively). The results show that the calibration models by NIRS are reliable and can be applied for the quantification for seven parameters from A. paniculata for quality control in traditional Chinese medicine production and processing.