偏最小二乘回归用于近红外光谱分析的稳健策略

偏最小二乘法（PLS）在近红外光谱（NIR）定量分析中占有重要地位,但预测结果往往容易受到样本分组和奇异样本等因素的影响,稳健性不强。多模型PLS (EPLS）方法在模型稳健性上得到提高,然而它无法识别样本中存在的奇异样本。为了同时提高模型的预测准确性和稳健性,本文提出了一种根据取样概率重新取样的多模型PLS方法,称为稳健共识PLS（RE-PLS）方法。该方法通过迭代赋权偏最小二乘法(IRPLS)计算样本回归残差得到每个校正集样本的取样概率,然后根据样本的取样概率来选择训练子集建立多个PLS模型,最后将所有PLS模型的预测结果平均作为最终预测结果。该方法用于两种不同植物样品的近红外光谱建模,并与传统的PLS及EPLS方法进行比较。结果表明该方法可以有效的避免校正集中奇异样本对模型的影响,同时可以提高预测精确度和稳健性。对于含有较多奇异样本的,复杂近红外光谱烟草实际样本,利用简单PLS或者EPLS方法建模预测效果不是很理想,而RE-PLS凭借其独特优势则有望在这种复杂光谱定量分析中得到广泛的应用。     

互信息诱导子空间集成偏最小二乘在近红外光谱定量校正中的应用

在集成框架下,提出了一种联合自助采样和基于互信息变量选择的子空间回归集成偏最小二乘算法MISEPLS.此算法的核心是通过训练集自助采样和随后计算互信息的方式来引入成员模型的差异性.由于互信息量小于一个特定阈值的变量被淘汰,每个成员模型在原始变量的一个子空间得到训练.模型融合考虑了简单平均和加权平均两种方式.通过两个近红外光谱定量校正实验,与建立单模型的全谱偏最小二乘算法(PLS)和基于互信息变量选择的偏最小二乘算法(MIPLS)进行了比较.结果表明,在不增加模型复杂度的情况下,MISEPLS能建立起更精确、更稳健的校正模型.     

X-射线荧光光谱分析中偏最小二乘回归技术与基本参数法的研究

将偏最小二乘回归技术(PLS)与X-射线荧光分析中的基本参数法(FP)相结合,编制PLSFP软件,运用熔样技术,可以定量测定地质岩石样品中的十三个主,次量元素,并使准确度提高,模型预测能力增强。模型选择和最佳维数的确定方法是成功的关键。PLSFP软件采用结构化编程,由标准FORTRAN 77语言写成,可在UNIX,XENIX和DOS操作系统下运行。     

Near-Infrared Spectroscopy Analytical Model Using Ensemble Partial Least Squares Regression

A novel ensemble-based feature selection method was developed which is designated as ensemble partial least squares regression coeffientents (EPRC). It was composed of two steps: generating a series of different single feature selectors and aggregating them to reach a consensus. Specifically, the bootstrap resampling approach was used to generate a diversity of single feature selectors, and the absolute values of the regression coefficients of the partial least squares (PLS) model were used to rank the features. Next, these feature rankings out of single feature selectors were aggregated by the weighted-sum approach. Finally, coupled with the regression model, the features selected by EPRC were evaluated through cross validation and an independent test set. By experiments of constructing the spectroscopy analysis model on three near infrared spectroscopy (NIRS) datasets, it was shown that the EPRC located key wavelengths, gave a promotion to regression performance, and was more stable and interpretable to the domain experts.     

基于多模型共识的偏最小二乘法用于近红外光谱定量分析

建立了多模型共识偏最小二乘(cPLS)建模方法, 并应用于烟草样品近红外(NIR)光谱与常规成分氯含量之间的建模研究, 探讨了建模参数对预测结果的影响. 结果表明, cPLS方法与传统的偏最小二乘算法(PLS)相比, 所建模型更稳定可靠, 预测结果也可得到了明显改善.     

血糖近红外光谱分析的Savitzky-Golay平滑模式与偏最小二乘法因子数的联合优选

利用偏最小二乘法(PLS)和光谱Savitzky-Golay(SG)平滑方法,建立血清葡萄糖近红外光谱分析的优化模型。基于最优单波数模型的预测效果,提出划分校正集和验证集的一种新方法。采用10000～5300cm-1和4920～4160cm-1的组合波段,光谱经过SG平滑处理,利用PLS方法建立定标预测模型。将平滑点数扩充为5,7,…,87(奇数),多项式次数扩充为n=2,3,4,5,6,得到包含582个平滑模式的14个平滑系数表。对所有平滑模式和PLS因子数(1～40)分别建立PLS模型。按照预测效果进行优选,得到最优SG平滑模式为1阶导数平滑,3、4次多项式类型,SG平滑点数为53,最优PLS因子数为7,最优RMSEP达到0.376mmol/L。所采用的划分校正集和验证集的方法、SG平滑模式的扩充、SG平滑模式和PLS因子数的联合大范围筛选能够有效地应用于近红外光谱分析的模型优化。     

样条变换集成罚函数偏最小二乘方法用于光谱数据重构和定量分析

针对高维小样本光谱数据所显现的函数型数据(Functional data)特性、与性质参数的非线性关系及变量间存有的严重共线性,采用了样条变换集成罚函数偏最小二乘回归新技术.它首先以三次B基样条变换实现非线性光谱数据的线性化重构,随后将重构的新光谱矩阵交由罚函数偏最小二乘法(Penalized PLS)构建其与性质参变量间的校正模型,其中罚函数中的光滑因子由交叉验证优化确定以调控模型的拟合精度.最后,通过小麦样品水分含量的近红外光谱定量分析,结果显示该技术光谱数据重构稳健,去噪明显,并有效解决高维小样本的过拟合和变量间的共线性,而预测集的均方根误差(RMSEP)为0.1808%,方法的非线性校正模型预测能力得到了明显提高.     

基于双脉冲LIBS结合偏最小二乘回归的稀土元素定量分析

激光诱导击穿光谱(LIBS)是一种以激光为激发源的等离子体发射光谱分析技术,已有将其用于稀土元素的定量分析研究,但由于稀土矿基体差异大、元素含量低,定量分析灵敏度和准确度仍有待提高。通过使用单激光分束构造双脉冲LIBS系统,并结合偏最小二乘回归(PLSR)算法实现对稀土矿石样品中的稀土元素La、Dy、Yb和Y的定量分析。结果表明,双脉冲LIBS结合PLSR可建立更加稳定的定标模型,与常规基本定标法相比,La、Dy、Yb和Y元素的相对均方根预测误差(RMSEP)从0.0061 %、0.0037%、0.0045%、0.0280 %降低至0.0044%、0.0016%、0.0029%、0.0134%,平均相对预测误差(AREP)从10.88%、15.27%、6.42%、17.20%降低至6.67%、3.62%、4.10%、7.98%。因此,双脉冲LIBS结合PLSR方法可以有效地提高LIBS对稀土矿石中稀土元素的定量分析能力。     

Comparison of Artificial Neural Networks with Partial Least Squares Regression for Simultaneous Determinations by ICP-AES

Simultaneous determination of several elements （U, Ta, Mn, Zr and W） with inductively coupled plasma atomic emission spectrometry （ICP-AES） in the presence of spectral interference was performed using chemometrics methods. True comparison between artificial neural network （ANN） and partial least squares regression （PLS） for simultaneous determination in different degrees of overlap was investigated. The emission spectra were recorded at uranium analytical line （263.553 nm） with a 0.06 nm spectral window by ICP-AES. Principal component analysis was applied to data and scores on 5 dominant principal components were subjected to ANN. A 5-5-5 （input, hidden and output neurons） network was used with linear transfer function after both hidden and output layers. The PI,S model was trained with five latent variables and 20 samples in calibration set. The relative errors of predictions （REP） in test set were 3.75% and 3.56% for ANN and PLS respectively.     

偏最小二乘法及主组分回归法用于药物组分的测定

本文研究了多元校准方法——偏最小二乘法(PLS)和主组份回归法(PCR)在药物多组份光度分析中的应用,获得了较满意的结果。而且在系列校准样品的实验设计、交叉证实法确定最佳因子数以及空缺组份体系的分析等方面进行了探讨。     

组合偏最小二乘回归方法在近红外光谱定量分析中的应用

针对近红外光谱数据局部效应显著,变量个数多,彼此间常存在严重的复共线性,并多与样品组分含量呈非线性关系,构建一种组合非线性偏最小二乘回归(E-S-QPLSR)方法。它采用无重复采样技术(subag-ging),从训练样本中生成若干子样,然后每个子样通过二次多项式偏最小二乘回归(QPLSR),建立其子模型,并实现对训练样本因变量的定量预测,再将它们交由线性PLS算法用于计算各子模型的组合权系数。将该法应用于80个玉米样品的水组分含量与其近红外光谱的定量关系建模,效果良好,显示出很强的学习能力,所建模型的预报性能也优于其它方法。     

近红外光谱结合偏最小二乘判别分析区分造纸法再造烟叶产品

本文采集162个造纸法再造烟叶产品的近红外光谱,结合偏最小二乘判别分析(PLS-DA)建立了再造烟叶产品的分类模型,实现了不同牌号再造烟叶产品的快速分类,并对45个预测集样品的牌号进行了分类预测。所建模型对校正集和预测集的预测正确率分别为100.0%和95.5%,与主成分分析(PCA)相比,PLS-DA对不同牌号再造烟叶产品的分类具有更好的效果。该模型为不同牌号再造烟叶产品分类提供了一种新的快速鉴别分析的方法,同时可初步监测再造烟叶产品的质量稳定性。     

Prediction of Properties of Soil Humic Substances from FTIR Spectra using Partial Least Squares Regression

Abstract

Humic substances (HS) play a key role in aquatic and terrestrial ecosystems. The understanding of the ecological functionalities of HS is based on the analysis of their properties, which is normally a very time-consuming procedure. Therefore we tested the possibility to apply the partial least squares regression (PLSR) method in connection with the mid infrared Fourier transform (FTIR) spectra of a series of soil humic substances for the prediction of different HS properties. The results with humic acid (HA) and fulvic acid (FA) fractions of soil HS from different environments show the possibility to predict several properties of an unknown soil HA with satisfying reliability above all the elemental composition.     

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。     

Chemometric Assisted UV-Spectrophotometric Methods Using Multivariate Curve Resolution Alternating Least Squares and Partial Least Squares Regression for Determination of Beta-Antagonists in Formulated Products: Evaluation of the Ecological Impact

In this study, UV-spectrophotometry coupled with chemometrics has been utilized to enhance the sustainability of quality control analysis of beta antagonists. First, we developed and optimized two eco-friendly chemometric-assisted methods without preliminary separation utilizing (1) multivariate curve resolution alternating least squares (MCR-ALS) and (2) well-established partial least squares regression (PLSR) multivariate calibration for the resolution and quantification of the most commonly prescribed beta antagonists in active pharmaceutical ingredients or commercial pharmaceutical products. The performance of the two proposed chemometric methods was computed and compared. Second, a comprehensive qualitative and quantitative evaluation of the eco-friendliness of the developed methods was performed utilizing the following greenness assessment tools: Green Analytical Procedure Index (GAPI), Analytical Eco-scale assessment (AES) tool, Raynie and Driver’s assessment tool and Analytical GREEnness Metric (AGREE). The models showed satisfactory recovery with a range from 99.83% to 101.12% for MCR-ALS and from 99.66% to 101.54% for PLSR. The optimized models were employed for green analysis of the investigated beta-blockers in single or co-formulated formulations without prior separation. The predictivity of the proposed MCR-ALS and the well-established PLSR method were very comparable. Nevertheless, the MCR-ALS method has the ability to recover the pure spectra of the studied analytes and the interferences as well. The proposed chemometric methods are fast, precise and do not need any sample pretreatment. In addition, they can be used as a benign substitute for the traditional methods used for the analysis of the investigated drugs in pharmaceutical products without harmful impacts on human health and the environment. They also provide advantages in terms of low solvent usage, reduced energy consumption and short analysis time, making them a safe and sustainable approach for quality control analysis.     

红外光谱定量分析中的一种变量聚类偏最小二乘算法

偏最小二乘算法( Partial least squares, PLS)可以很好地解决分析数据中的变量共线性问题,在光谱分析,尤其是近/中红外及拉曼光谱的定量分析中应用广泛。针对PLS存在的有效信息提取和噪声抑制问题,提出一种变量聚类重加权的PLS算法。通过对光谱的各波数变量进行聚类并分别建模,然后集成为全谱模型。通过对计算并赋予各子类不同的权重,根据对模型的贡献对变量进行重加权,从而提高算法的预测精度。汽油中的辛烷值预测和烟草中的烟碱含量预测两组近红外数据验证表明,所提出算法优于经典的PLS算法,其RMSEP在两组数据中分别降低32%和22%,在光谱数据的定量分析中具有潜在的应用优势。     

偏最小二乘变量筛选法在毒品来源分析中的应用

提出了一种新的偏最小二乘变量筛选方法,该方法利用PLS回归建模过程中的一些信息,删除一部分冗余的或对建模影响不大的变量来简化、优化预报模型。用此方法结合变量扩维方法处理云南昆明、思茅、西双版纳3个来源地缴获的244个海洛因样本的ICP-MS数据时,与传统的算法比较,模型的判别准确率得到大大提高,达到95%以上。且所得到的模型含变量少,很容易分析或解释各变量对模型的影响。因此该方法可用于对毒品来源有效的识别或鉴定。