苹果产地差异对可溶性固形物近红外光谱检测模型影响的研究

为更好地利用近红外光谱预测苹果可溶性固形物含量,减少产地差异对近红外光谱检测模型的影响,以4种不同产地的富士苹果为研究对象,采用基于x-y共生距离的样本划分方法分别对不同产地的苹果选取代表性样本作为校正集,利用偏最小二乘算法,建立和比较单一产地和混合产地下的苹果可溶性固形物近红外光谱检测模型,并结合竞争性自适应重加权算法(CARS)和连续投影算法(SPA)对苹果可溶性固形物的建模变量进行筛选。相比单一产地和其它混合产地模型,混合所有4种苹果产地的校正集样本建立的模型取得了最好的预测结果,另外,结合CARS-SPA筛选的16个特征波长,模型得到了进一步简化,其预测相关系数和预测均方根误差分别为0.978和0.441°Brix。结果表明,利用多个产地的苹果样本建立的混合模型,结合有效特征波长,可提高对苹果可溶性固形物含量的预测精度,减小产地差异对可溶性固形物近红外光谱检测的影响。     

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。     

自适应蚁群优化算法的近红外光谱特征波长选择方法

为提高近红外光谱预测模型的精度和适用性,同时简化模型,提出了自适应蚁群优化偏最小二乘法优选特征波长的方法,建立不同产地苹果可溶性固形物含量混合分析模型。收集山东、陕西和新疆的富士苹果,采集3800~14000 cm"1范围的近红外光谱,并对其重要品质指标可溶性固形物含量进行测定。利用蚁群算法启发式全局搜索的特点,结合蒙特卡罗轮盘赌随机选择机制,优选苹果可溶性固形物含量的近红外光谱特征波长,然后用偏最小二乘法建立分析模型。与全光谱偏最小二乘模型和遗传偏最小二乘模型相比,蚁群优化算法选择的波长数最少,模型预测能力最强,预测的相关系数R和预测均方根误差RMSEP分别为0.9708和0.5144。研究结果表明,自适应蚁群优化算法可以有效选择近红外光谱特征波长,提高模型的稳健性和适用性。     

中红外光谱技术应用于饶河蜂蜜产地溯源的表征

通过对饶河本地的蜂蜜样品和其它地区蜂蜜样品的红外光谱谱图进行分析,利用化学计量软件建立饶河黑蜂蜂蜜产地真假判别模型。利用中红外光谱分析仪,分别采用5倍稀释和10倍稀释的前处理方法对95个饶河本地蜂蜜样品和其它地域具有代表性的31个蜂蜜样品进行测定,利用质量控制模块和分析软件对样品集和校正集进行建模处理,对建立的模型加以筛选和优化,最终确定主成分数F=3、门槛值T=2所建立的判别模型符合建模要求,该模型可以对饶河黑蜂蜂蜜产地进行较为准确的判定,判定准确率为90.3%。结果表明中红外光谱技术对饶河黑蜂蜂蜜进行真伪判别是一种有效的方法。     

近红外光谱法快速鉴定油菜杂交种的纯度

建立近红外光谱技术测定油菜杂交种纯度的方法。考察了样品杯类型、光谱预处理方法和波长范围对近红外模型预测性能的影响。结果发现,由不同样品杯采集近红外光谱所建立的校正模型,其预测性能存在较大的差异,旋转杯明显优于安瓿瓶;采用消除常数偏移量对光谱进行预处理能有效地提取光谱信息,选择5 000~8 000 cm–1波数范围作为建模谱区,其包含的有效信息率最高。在最佳条件下建立油菜杂交种纯度的校正模型,其决定系数(R2)为0.980 0,交互验证均方根误差(RMSECV)为0.008 59。利用该模型对预测集进行测定,预期均方根误差(RMSEP)为0.007 59,表明该模型具有很好的预测性能,近红外光谱法用于杂交种纯度的鉴定是可行的。     

烟草组分的近红外光谱和支持向量机分析

测定了120个产自福建、安徽和云南烟草样品的近红外光谱. 在利用支持向量机(SVM)技术建立其定量、定性分析模型之前, 用小波变换技术对光谱变量进行了有效的压缩, 然后采用径向基核函数建立了75个烟草样品的分类模型, 同时建立了总糖、还原糖、烟碱和总氮4个组分的定量分析模型, 并利用45个烟草样品对模型进行了检验. 仿真实验表明, 建立的SVM分类模型分类准确率达到100%, 而4个组分的定量分析模型的预测决定系数(R2)、预测均方差(RMSEP)和平均相对误差(RME)3个指标值显示其模型泛化能力非常强, 预测效果良好, 可见这是一种有效的近红外光谱的建模分析方法.     

近红外光谱结合LSTSVM算法判别不同等级烟叶

本文用近红外光谱结合最小二乘双胞胎支持向量机(LSTSVM)算法建立了烟叶等级分类模型。从三个等级共210个烟叶样品中,取出120个样品作为建模集,剩余90个样品作为预测集。为了建立最优模型,对光谱预处理方法和模型参数进行筛选优化,最优模型对预测集样品的平均识别率为95.56%,结果表明该方法可以作为烟叶等级分类的一种有效方法。此外,将该算法与SIMCA、PLS-DA、SVM等三种常见的模式识别算法进行了比较,结果表明基于样品的原始光谱,同等条件下,LSTSVM算法的预测效果优于其他三种算法。     

近红外反射光谱法分析玉米秸秆纤维素含量的研究

利用近红外反射光谱分析技术和偏最小二乘回归法（PLS）,通过比较不同光谱范围和光谱预处理方法,采用二阶导数光谱预处理,在7540．3-5361．1cm^-1和4882.9—4504．9cm^-1谱区内建立了近红外光谱测定玉米秸秆纤维素含量的校正模型。利用15个玉米秸秆样品对所建模型的实际预测效果进行了验证,预测值与化学值的相关系数（r）可达0．9953,最大相对误差仅为5．20。结果表明,近红外光谱技术可以快速、准确地测定玉米秸秆纤维素,该结果对玉米秸秆材料的快速鉴定和筛选利用具有重要的意义。     

基于稀疏主成分分析特征选择算法的山楂叶产地判别模型研究

为实现山楂叶产地的快速判别,提出一种基于稀疏主成分分析特征选择（SPCAFS）与支持向量机（SVM）建模的定性分析方法。采用近红外积分球漫反射光谱法采集6个产地共41批山楂叶123份样品的近红外光谱图,经数据预处理后,通过SPCAFS对代表性特征波段进行选择,并采用SVM建立山楂叶近红外产地判别模型。模型与连续投影（SPA）,正则化自表示（RSR）和稀疏子空间聚类（SSC）3种特征选择算法进行对比,以准确率、精确度和灵敏度作为评价标准,评估所提模型的预测性能。结果显示,SPCAFS的特征波段数相比于全波长建模从1 500减少到21,预测结果的准确率和精确度分别从78%、76%提升至97%、100%。同时,相比于SPA、RSR、SSC算法,准确率分别提升了6%、3%、3%,精确度分别提升了13%、10%、5%,模型的预测能力得到显著提升,基于SPCAFS的SVM判别模型可实现山楂叶南北产地的快速判别。     

遗传算法用于偏最小二乘方法建模中的变量筛选

利用全局搜索方法－遗传算法（genetic algorithms,GA)对近红外光谱分析中的波长变量进行筛选,再用偏最小二乘方法（patrial least squares,PLS）建立分析校正模型。对两类样品的近红外光谱分析应用实例表明,这种选取变量进行校正的方法,不仅简化、优化了模型,而且增强了所建模型的预测能力,尤其适用于单纯PLS较以校正关联的体系。     

废旧涤/棉混纺织物近红外定量分析模型的建立及预测

将中红外光谱筛选出的598个纯涤、纯棉及涤/棉混纺样本采用GB/T 2910.11-2009法测定其涤、棉准确含量,其中校正集样本252个,验证集样本346个。使用便携式近红外光谱仪获取样本的原始近红外光谱(NIRS)。校正集样本依据回归系数的分布趋势和范围选取最佳建模谱区,并采用差分一阶导、S-G平滑和均值中心化相结合的方法对原始光谱进行预处理,利用偏最小二乘法(PLS)建立涤/棉混纺织物中涤含量的近红外(NIR)定量分析模型。同时分析了样本颜色对NIRS的影响,探讨了斜线光谱样本、奇异样本和不同组织结构织物对模型预测效果的影响。结果表明:利用PLS法建立的涤/棉混纺织物定量分析模型最优组合包含1个光谱区间和9个主成分因子,校正集相关系数(RC)为0.998,标准偏差(SEC)为0.908。为验证所建模型的有效性和实用性,对346个未参与建模的涤棉样本进行了预测,并将预测结果与国标法测定值进行方差分析,两种方法结果无显著差异,预测正确率达97%以上。模型的建立为废旧涤/棉混纺织物快速、无损分拣提供了基础数据库。     

Application of Grey-correlated Spectral Region Selection in Analysis of Near-infrared Spectra

The optimal selection method of spectral region based on the grey correlation analysis was applied in the analysis of near-infrared(NIR) spectra. In order to compute “characteristic” spectral region, 160 samples of tobacco were surveyed by NIR. Next, the whole spectral region was randomly divided into six regions, and the values of association coefficients and correlation orders of different regions were computed for total sugar, reducing sugar and nicotine. Moreover, two regions that owned the largest value of association coefficient were regarded as “characteristic” spectral region of a model. Finally, the quantitative analysis models of different components were established via the partial least squares method, and the common selection methods of spectral region were compared. The simulation results indicate that the models to choose the spectral region based on grey correlation analysis are more effective than the common selection methods of spectral region, the optimized time of algorithm is shorter, the prediction precision of the models is higher and generalization ability for quantitative analysis results is stronger. This research can provide the support for the quantitative analysis models of NIR spectra and new idea for commercial analysis software of NIR. So, it has a high application value in the analysis of NIR spectra.     

刺五加注射液近红外含量预测模型谱段选择规律和消除溶剂干扰方法的探讨

目前中药制剂生产过程中缺乏全过程参数检测和质量控制技术手段,不同生产批次药品化学成分差异较大、质量不够稳定、临床使用疗效和安全性不理想,因此,建立其完善的质量评价体系及其准确快速的质量评价方法,成为中药质量控制的重中之重.通过对刺五加注射液近红外(near infrared,NIR)含量预测模型的谱段选择规律和消除溶剂干扰方法的探讨,发现采用表征混合物结构差异的结构相关谱段结合含量相关谱段作为NIR组分预测模型谱段,用基于水为参比光谱的样本光谱建立含量预测模型,并利用水作为参比光谱识别和提取待分析组分的光谱信息,可以得到较理想的NIR含量预测结果.通过对刺五加注射液中绿原酸、紫丁香苷和刺五加苷E组分的近红外光谱结构相关谱段和含量相关谱段的归属,分别建立了绿原酸、紫丁香苷和刺五加苷E组分的含量预测模型,可用于快速分析刺五加注射液中不同组分的含量.     

A procedure for calibration transfer between near-infrared instruments--a worked example using a transmittance single tablet assay for piroxicam in intact tablets

A procedure was developed for different modes of calibration transfer in near-infrared (NIR) spectroscopy, which included a method for the selection of a subset of samples appropriate for transfer. As a worked example, these guidelines were applied to the transfer of a multivariate calibration model, representing a validated NIR single tablet assay for the active within an intact pharmaceutical product, between three equivalent dispersive NIR transmission instruments. Transfer was first evaluated between two instruments, representing the situation where both were available during calibration development. A spectral correction method alone, applied to the transfer instrument, was not sufficient to facilitate transfer, with further optimisation of the calibration model using a novel wavelength selection algorithm necessary to remove regions of the spectral range that resulted in skewed predictions on the second instrument. Through this approach, a single calibration model was found to be equally accurate and precise on the two instruments. A procedure, using the Kennard-Stone algorithm, is described for determining a reduced number of samples as a transfer set using only the spectral information from the original instrument. The purpose of the subset was to permit transfer to a new instrument where that instrument was not available until after calibration development or where it was undesirable to re-measure the full sample set (i.e. due to excessive reference chemistry). Utilising the transfer set, transfer to a third instrument was evaluated. The calibration model, optimised between the first two instruments, was not directly applicable for the third instrument, with further wavelength selection required to remove a small region of spectral data. On completion, using a full statistical evaluation, a single calibration model was found to be equally accurate and precise on all three instruments.     

近红外光谱技术结合主成分分析法用于子宫内膜癌的诊断

应用近红外光谱技术结合化学计量学方法研究了子宫内膜癌组织近红外光谱特征提取和早期诊断的可行性. 测定了154 例子宫内膜组织切片的近红外光谱, 选取适宜的波段和光谱预处理方法进行主成分分析, 很好地区分了癌变、增生和正常子宫内膜组织切片, 并且分辨出处于不同分化期的组织切片, 为子宫内膜癌的早期诊断提供了可靠依据. 该法快速、简便, 有望发展成为一种新型的肿瘤无创诊断方法.     

Rapid and Nondestructive Analysis of Bacillus Calmette–Guerin Polysaccharide Nucleic Acid Injection by near-Infrared Spectroscopy with Chemometrics

A novel quantitative analytical method using near-infrared (NIR) spectroscopy combined with chemometrics has been developed to determine the polysaccharides and nucleic acids for routine quality analysis of Bacillus Calmette–Guerin polysaccharide and nucleic acid injections. A Monte-Carlo method was used to detect and discard outliers and to improve the predictive ability of the model. Various other spectral preprocessing methods such as smoothing, derivative, multiplicative scattering correction, standard normal variables, and orthogonal signal correction methods were used to remove noise and other irrelevant information from the spectra. Sample-set partitioning based on joint x–y distance method was utilized to divide the sample measurements into calibration and validation datasets. The optimal wavelength variables were determined by competitive adaptive weighted sampling. The model was established and cross-validated using partial least square regression. The root mean square errors of cross-validation for polysaccharides and nucleic acids were determined to be 0.0382 and 5.218, and the root mean square errors of prediction were 0.0229 and 6.282. The overall results show that NIR spectroscopy combined with chemometry is effective for the quantitative analysis of Bacillus Calmette–Guerin polysaccharide and nucleic acid injections.     

最小角回归结合核极限学习机的近红外光谱对柑橘黄龙病的鉴别

传统的柑橘黄龙病检测方法存在准确度低、稳定性差等问题,该文提出了一种基于最小角回归结合核极限学习机(Least angle regression combined with kernel extreme learning machine,LAR-KELM_((RBF)))的近红外柑橘黄龙病鉴别方法。该方法将光谱数据通过小波变换进行预处理,然后用最小角回归(LAR)算法进行光谱波长的筛选,最后通过核极限学习机(KELM_((RBF)))实现样本的分类。实验采用柑橘叶片的近红外光谱数据,验证了LAR-KELM_((RBF))算法的性能,其分类准确度最高为99.91%,标准偏差为0.11。不同规模训练集的实验结果表明,LAR-KELM_((RBF))模型较极限学习机(ELM)、波形叠加极限学习机(SWELM)、反向传播神经网络(BP_((2层)))、KELM_((RBF))和支持向量机(SVM)模型分类准确度高、稳定性强,能够广泛应用于柑橘黄龙病的检测鉴别。     

Determination of chlorogenic acid in plant samples by using near-infrared spectrum with wavelet transform preprocessing.

By theoretical analysis, it is found that wavelet transform (WT) with a wavelet function can be regarded as a smoothing and a differentiation process, and that the order of differentiation is determined by the vanishing moment, which is an important property of a wavelet function. Therefore, a method based on the continuous wavelet transform (CWT) for removing the background in the near-infrared (NIR) spectrum is proposed, and it is used in the determination of the chlorogenic acid in plant samples as a preprocessing tool for partial least square (PLS) modeling. It is shown that the benefit of the proposed method lies not only in its performance to improve the quality of PLS model and the prediction precision, but also in its simplicity and practicability. It may become a convenient and efficient tool for preprocessing NIR spectral data sets in multivariate calibration.     

基于灰狼算法的近红外光谱变量选择方法研究

基于群体智能的灰狼优化(GWO)算法具有参数少、结构简单、易于实现的优点,但在光谱领域的应用较少。该研究将GWO算法引入近红外光谱的变量筛选中,以玉米数据为例,考察了GWO算法中狼群性能、迭代次数、狼群数量及运算效率,并建立了偏最小二乘(PLS)模型对玉米样品中蛋白质、脂肪、水分以及淀粉含量的测定。结果显示,GWO算法运算效率很高,经过参数调优后建立PLS模型,其蛋白质、脂肪、水分及淀粉的保留变量数分别为19、19、14、34,预测均方根误差(RMSEP)从全波长PLS建模的0.245 8、0.122 4、0.339 8、1.105 8分别下降到0.147 7、0.080 1、0.176 2、0.739 8,分别下降了40%、35%、48%、33%,相关系数也相应地提高。因此,GWO算法不仅优化速度快,选择变量数少,还可以显著提高PLS模型的预测精度,是一种近红外光谱变量选择的有效方法。     

An adaptive strategy for selecting representative calibration samples in the continuous wavelet domain for near-infrared spectral analysis

Sample selection is often used to improve the cost-effectiveness of near-infrared (NIR) spectral analysis. When raw NIR spectra are used, however, it is not easy to select appropriate samples, because of background interference and noise. In this paper, a novel adaptive strategy based on selection of representative NIR spectra in the continuous wavelet transform (CWT) domain is described. After pretreatment with the CWT, an extension of the Kennard–Stone (EKS) algorithm was used to adaptively select the most representative NIR spectra, which were then submitted to expensive chemical measurement and multivariate calibration. With the samples selected, a PLS model was finally built for prediction. It is of great interest to find that selection of representative samples in the CWT domain, rather than raw spectra, not only effectively eliminates background interference and noise but also further reduces the number of samples required for a good calibration, resulting in a high-quality regression model that is similar to the model obtained by use of all the samples. The results indicate that the proposed method can effectively enhance the cost-effectiveness of NIR spectral analysis. The strategy proposed here can also be applied to different analytical data for multivariate calibration.