同时测定水溶液中葡萄糖、果糖和蔗糖的近红外光谱法

通过对27个葡萄糖、果糖和蔗糖水溶液的混合体系进行近红外光谱分析，建立了葡萄糖、果糖和蔗糖含量测定的偏最小二乘法(PLS)模型；葡萄糖、果糖和蔗糖的线性范围分别为0—300g/L、0—200g/L、0—300g/L，模型校正集的标准误差(SEC)分别为1．4、1．8、1．4g/L；用该模型对6个样品进行分析，葡萄糖、果糖和蔗糖含量测定的相对标准偏差(RSD)分别为1．2％、2．6％和1．8％。     

遗传算法在近红外无创伤人体血糖浓度测量基础研究中的应用

遗传算法（GA）应用在偏最小二乘法（PLS）校正模型的波长优化选择中具有显著的效果。将遗传算法作为模块循环运行，能更快达到最优解，有效提高测量精度，减少建模所用波长数。本文将该方法应用于无创伤人体血糖浓度光学检测的基础研究中，验证实验所用样品为：①葡萄糖水溶液；②包含牛血红蛋白和白蛋白的葡萄糖水溶液；③人血中的血浆（含葡萄糖）。结果表明：建模的波长个数可分别减少88%、86%、85%；预测标准偏差（RMSEP）分别减少56%、64%。这对无阶伤人体血糖浓度光学检测理论的进一步研究具有指导意义。     

浓度分段模型应用于人体血清葡萄糖的FTIR/ATR光谱分析

利用傅立叶变换红外光谱(FTIR)和衰减全反射(ATR)技术,建立了人血清葡萄糖的快速定量分析方法。根据葡萄糖水溶液与纯净水差谱得到葡萄糖的指纹吸收波段(1 200~900 cm-1),分别在全谱(4 000~600 cm-1)和指纹波段建立偏最小二乘法(PLS)模型,指纹波段的预测效果明显好于全谱。选择指纹波段后,提出一种根据浓度分段分别建模然后进行组合的建模方法。按照全部样品、低浓度样品、高浓度样品分别建立模型后,根据3个模型进行综合决策。应用独立的检验集对样品进行测试表明,按葡萄糖浓度范围分段建立组合模型的预测效果优于基于全部样品建模的预测效果。对于分段阈值附近的样本,低浓度和高浓度模型的预测效果差别不大。浓度分段组合模型的预测均方根偏差(RMSEP)和预测相关系数(Rp)分别为0.732mmol/L和0.948。     

浓度分段模型应用于人体血清葡萄糖的FTlR/ATR光谱分析

利用傅立叶变换红外光谱（ FTlR）和衰减全反射（ ATR）技术,建立了人血清葡萄糖的快速定量分析方法。根据葡萄糖水溶液与纯净水差谱得到葡萄糖的指纹吸收波段（1200~900 cm-1）,分别在全谱（4000~600 cm-1）和指纹波段建立偏最小二乘法（ PLS）模型,指纹波段的预测效果明显好于全谱。选择指纹波段后,提出一种根据浓度分段分别建模然后进行组合的建模方法。按照全部样品、低浓度样品、高浓度样品分别建立模型后,根据3个模型进行综合决策。应用独立的检验集对样品进行测试表明,按葡萄糖浓度范围分段建立组合模型的预测效果优于基于全部样品建模的预测效果。对于分段阈值附近的样本,低浓度和高浓度模型的预测效果差别不大。浓度分段组合模型的预测均方根偏差（ RMSEP）和预测相关系数（ Rp ）分别为0.732 mmol/L和0.948。     

近红外光谱法定量分析水溶液中葡萄糖、麦芽糖和β-环糊精

本文应用正交设计法配制了葡萄糖、麦芽糖和伊环糊精混合水溶液的29个样品（其中25个用作校正集样品,4个用作预测集样品）,扫描此混合水溶液样品的近红外（NIR）光谱,并用Unscrambler定量分析软件将近红外光谱与对应的化学成分值相关联,在不同波长范围下建立了水溶液中葡萄糖、麦芽糖和p环糊精的近红外原始光谱、一阶导数光谱的偏最小二乘法（PLS）定量分析数学模型,根据模型评价参数选择了最佳模型,并且用这些模型对未知样品进行了预测。结果显示,对浓度范围分别在22.73-104.17mg·mL^-1、22.73-104.17mg·mL^-1和1.85-8.41mg·mL^-1的葡萄糖、麦芽糖和仔环糊精水溶液,其NIR数学模型的相关系数分别为0.9946、0.9976和0.9036,模型的预测相对偏差（RMSEP）分别为2.52mg·mL^-1、1.95mg·mL^-1和0.87mg·mL^-1,预测标准偏差（SEP）分另4为2.58mg·mL^-1、2.01mg·mL^-1和0.90mg·mL^-1。     

基于近红外技术快速无损分析整粒棉籽中的脂肪酸含量

应用近红外光谱技术可以实现整粒带壳作物种子中脂肪酸含量的快速、无损分析。以385份棉花种子为实验材料,应用线性的偏最小二乘(PLS)和非线性的最小二乘支持向量机(LS-SVM)方法,结合蒙特卡罗无信息变量消除法(MC-UVE),构建整粒棉籽中脂肪酸含量的近红外校正模型。结果表明,基于变量选择的LS-SVM模型具有最佳的预测性能,其棕榈酸、硬脂酸、油酸、亚油酸、饱和脂肪酸和不饱和脂肪酸含量的近红外校正模型的相关系数R2分别为0.863,0.881,0.843,0.806,0.894和0.917,剩余预测偏差RPD分别为2.669,2.880,2.508,2.202,3.023和3.473。本方法省略了种子的粉碎过程,MC-UVE方法有助于提高校正模型的稳健性和精确度。     

偏最小二乘法测定复方乙酰水杨酸片中的有效成分

将偏最小二乘法(PLS)与近红外漫反射光谱法相结合,对复方乙酰水杨酸片进行无损非破坏定量分析．建立了最佳的数学校正模型,比较了样品中3种有效成分(乙酰水杨酸、非那西丁和咖啡因)同时测定和单独测定时的主成分数对PLS定量预测能力的影响,预测了未知样品。3种有效成分同时测定和单独测定建立的PLS模型具有相同的主成分数,PLS预报浓度与参考浓度具有相近的标准偏差,说明用PLS法同时测定3种组分的含量是可行的。     

径向基神经网络奥斯特杨方波伏安法同时测定铬和锌

径向基函数神经网络(RBFN)和核心偏最小二乘法(KPLS)用于分析重叠的Cr（Ⅲ）和Zn(Ⅱ）的奥斯特杨(Osteryoung)方波伏安图，程序SPRBFN和SPKPLS被设计用于全部计算。在RBFN方法中，普通高斯函数可用作隐藏层非线性转移函数。由于其局部性质，RBFN能被快速训练，避免陷入局部最小。对两个方法预测能力的研究结果显示其所有组分的相对预测标准偏差(RSEP)分别为0．677％和13．0％。因此，RBFN方法较之K眦方法可提供更为精确的结果，而且在解决局部最小，改进收敛速率方面也不失为一个重要的工具。     

近红外光谱法同时测定通天口服液中天麻素和芍药苷的含量

本文应用近红外光谱结合偏最小二乘法建立了同时测定通天口服液中天麻素与芍药苷含量的方法。以高效液相色谱(HPLC)法测定通天口服液样品中天麻素和芍药苷的化学参考值,随机抽取60个样本作校正集,20个样本作预测集。用偏最小二乘法(PLS)将校正集样本的近红外光谱与相应样本的天麻素和芍药苷含量分别相关联建立模型。结果表明,天麻素和芍药苷校正模型的决定系数分别为96.28%、94.55%,模型的交叉验证均方差分别为0.0336、0.00908,预测集的决定系数分别为94.23%、92.86%,预测集均方差分别为0.0453、0.00839。同时还做了模型的精密度实验,该方法能用于大批量样品的快速分析。     

复杂样品近红外光谱定量分析模型的构建方法

针对复杂样品近红外光谱分析中校正集的设计问题, 探讨了标准样品参与复杂样品建模的可行性. 通过标准样品和复杂基质样品共同构建的偏最小二乘(PLS)模型, 考察了波段筛选和建模参数对预测结果的影响. 结果表明, 采用PLS方法建立定量模型时, 校正集样品性质应该尽量与预测集样品相似, 当样品的性质相差较大时, 适当增加校正集样品的差异性可使模型具有更强的预测能力. 同时, 波段优选对提高预测结果的准确性具有重要的意义.     

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.     

Use of radial basis function networks and near-infrared spectroscopy for the determination of total nitrogen content in soils from Sao Paulo State

Total nitrogen has been determined by using a model developed between the conventional chemical measurements and diffuse reflectance spectra in the near-infrared region. Samples (244) from different types of soils with total nitrogen contents ranging from 0.20 to 13.60% (m/m) were modeled by partial least-squares regression (PLS), multi-layer perceptron feed-forward networks (MLP) and radial basis function networks (RBFN). The RBFN model produced a better square error of prediction (SEP) of 0.048 and R(2) = 0.93 in a procedure that is simpler, faster and less dependent on the initial conditions.     

Determination of Basic Nitrogen in Residues of Crude Oil Distillation Using ATR-FTIR and Chemometric Methods

A method for basic nitrogen determination in residues of crude oil distillation using infrared spectroscopy and chemometrics algorithms was developed. Interval partial least squares, synergy interval partial least squares, and backward interval partial least squares were evaluated for calibration model construction. The samples were divided into a calibration and prediction set containing 40 and 15 samples, respectively. The first derivative with a Savitzky-Golay filter and the mean centered data showed the best results and were used in all calibration models. The backward interval partial least squares algorithm with spectra divided in 60 intervals and combinations of 4 intervals (1407 to 1372; 1117 to 1082; 971 to 936; 914 to 879 cm^?1) showed the best root mean square error of prediction of 0.016 wt%. This calibration model displayed a suitable correlation coefficient between reference and predicted values.     

Second-order multivariate calibration procedures applied to high-performance liquid chromatography coupled to fast-scanning fluorescence detection for the determination of fluoroquinolones

Different second-order multivariate calibration algorithms, namely parallel factor analysis (PARAFAC), N-dimensional partial least-squares (N-PLS) and multivariate curve resolution-alternating least-squares (MCR-ALS) have been compared for the analysis of four fluoroquinolones in aqueous solutions, including some human urine samples (additional four fluoroquinolones were simultaneously determined by univariate calibration). Data were measured in a short time with a chromatographic system operating in the isocratic mode. The detection system consisted of a fast-scanning spectrofluorimeter, which allows one to obtain second-order data matrices containing the fluorescence intensity as a function of retention time and emission wavelength. The developed approach enabled us to determine eight analytes, some of them with overlapped profiles, without the necessity of applying an elution gradient, and thus significantly reducing both the experimental time and complexity. The study was employed for the discussion of the scopes of the applied second-order chemometric tools. The quality of the proposed technique coupled to each of the evaluated algorithms was assessed on the basis of the figures of merit for the determination of fluoroquinolones in the analyzed water and urine samples. Univariate calibration of four analytes led to limits of detection in the range 20–40 ng mL⁻¹ and root mean square errors for the validation samples in the range 30–60 ng mL⁻¹ (corresponding to relative prediction errors of 3–8%). The ranges for second-order multivariate calibration (using PARAFAC and N-PLS) of the remaining four analytes were: limit of detection, 2–8 ng mL⁻¹, root mean square errors, 3–50 ng mL⁻¹ and relative prediction errors, 1–5%.     

Evaluation of selectivity and robustness of near-infrared glucose measurements based on short-scan Fourier transform infrared interferograms

The selectivity and robustness of near-infrared (near-IR) calibration models based on short-scan Fourier transform (FT) infrared interferogram data are explored. The calibration methodology used in this work employs bandpass digital filters to reduce the frequency content of the interferogram data, followed by the use of partial least-squares (PLS) regression to build calibration models with the filtered interferogram signals. Combination region near-IR interferogram data are employed corresponding to physiological levels of glucose in an aqueous matrix containing variable levels of alanine, sodium ascorbate, sodium lactate, urea, and triacetin. A randomized design procedure is used to minimize correlations between the component concentrations and between the concentration of glucose and water. Because of the severe spectral overlap of the components, this sample matrix provides an excellent test of the ability of the calibration methodology to extract the glucose signature from the interferogram data. The robustness of the analysis is also studied by applying the calibration models to data collected outside of the time span of the data used to compute the models. A calibration model based on 52 samples collected over 4 days and employing two digital filters produces a standard error of calibration (SEC) of 0.36 mM glucose. The corresponding standard errors of prediction (SEP) for data collected on the 5th (18 samples) and 7th (10 samples) day are 0.42 and 0.48 mM, respectively. The interferogram segment used for the analysis contained only 155 points. These results are compatible with those obtained in a conventional analysis of absorbance spectra and serve to validate the viability of the interferogram-based calibration.     

Selection of the efficient wavelength regions in FT-NIR spectroscopy for determination of SSC of ‘Fuji’ apple based on BiPLS and FiPLS models

The feasibility that used the efficient selection of wavelength regions in FT-NIR for a rapid and conclusive determination of fruit inner qualities such as soluble solids content (SSC) of apples was investigated. An apples NIRS acquisition device was developed in this study. With this device, the apple was rolling while collecting the NIR spectroscopy. Graphically oriented local multivariate calibration modeling procedures such as interval partial least-squares (iPLS), backward interval partial least-squares (BiPLS), and forward interval partial least-squares (FiPLS) were applied to select the efficient spectral regions that provides the lowest prediction error, in comparison to the full-spectral model. Among 40 intervals, the optimal combinations of 10 spectral intervals were chosen by FiPLS to obtain a satisfactory result, while those of 5 by BiPLS for the simplicity. The intervals chosen by BiPLS are not the same as those by FiPLS, due to the different algorithm of the two methods. In the determinations, a root mean square error of prediction (RMSEP) of 0.732 was obtained after interval selection.     

Genetic Algorithm Applied to Selection of Wavelength in Partial Least Squares for Simultaneous Spectrophotometric Determination of Nitrophenol Isomers

Abstract

Ternary mixtures of nitrophenol isomers have been simultaneously determined in synthetic and real matrix by application of genetic algorithm and partial least squares model. All factors affecting the sensitivity were optimized and the linear dynamic range for determination of nitrophenol isomers found. The simultaneous determination of nitrophenol mixtures by using spectrophotometric methods is a difficult problem, due to spectral interferences. The partial least squares modeling was used for the multivariate calibration of the spectrophotometric data. A genetic algorithm is a suitable method for selecting wavelength for PLS calibration of mixtures with almost identical spectra without loss prediction capacity. The experimental calibration matrix was designed by measuring the absorbance over the range 300–520 nm for 21 samples of 1–20 µg mL^?1, 1–20 µg mL^?1, and 1–10 µg mL^?1 of m‐nitrophenol, o‐nitrophenol, and p‐nitrophenol, respectively. The root mean square error of prediction for m‐nitrophenol, o‐nitrophenol, and p‐nitrophenol with genetic algorithms and without genetic algorithms were 0.3732, 0.5997, 0.3181 and 0.7309, 0.9961, 1.0055, respectively. The proposed method was successfully applied for the determination of m‐nitrophenol, o‐nitrophenol, and p‐nitrophenol in synthetic and water samples.     

近红外光谱-径向基神经网络在异烟肼片无损定量分析中的应用

应用异烟肼片粉末的近红外漫反射光谱数据分别结合偏最小二乘法(PLS)和径向基神经网络(RBFNN)建立定量分析模型,并用所建模型对预测集样品进行了预测,结果表明:应用RBFNN所建立的定量分析模型优于PLS模型,相关系数(r)值由0.99593提高到0.99734,交互验证均方根误差(RMSECV)值由0.00523下降到0.00423,预测均方根误差(RMSEP)值由0.00614下降到0.00501。     

Quality assessment of gasoline using comprehensive two‐dimensional gas chromatography combined with unfolded partial least squares: A reliable approach for the detection of gasoline adulteration

Comprehensive two‐dimensional gas chromatography and flame ionization detection combined with unfolded‐partial least squares is proposed as a simple, fast and reliable method to assess the quality of gasoline and to detect its potential adulterants. The data for the calibration set are first baseline corrected using a two‐dimensional asymmetric least squares algorithm. The number of significant partial least squares components to build the model is determined using the minimum value of root‐mean square error of leave‐one out cross validation, which was 4. In this regard, blends of gasoline with kerosene, white spirit and paint thinner as frequently used adulterants are used to make calibration samples. Appropriate statistical parameters of regression coefficient of 0.996–0.998, root‐mean square error of prediction of 0.005–0.010 and relative error of prediction of 1.54–3.82% for the calibration set show the reliability of the developed method. In addition, the developed method is externally validated with three samples in validation set (with a relative error of prediction below 10.0%). Finally, to test the applicability of the proposed strategy for the analysis of real samples, five real gasoline samples collected from gas stations are used for this purpose and the gasoline proportions were in range of 70–85%. Also, the relative standard deviations were below 8.5% for different samples in the prediction set.