简化苹果糖度预测模型的近红外光谱预处理方法

采用正交信号校正法(OSC)和净分析物预处理法(NAP)分别对苹果的近红外光谱(1300~2100 nm)进行预处理,并结合偏最小二乘法(PLS)建立了糖度预测模型。应用结果显示,随着预处理过程中所用的正交信号校正因子或净分析物预处理因子的逐渐增加,偏最小二乘糖度模型(OSC/PLS模型和NAP/PLS模型)所采纳的最佳因子数也会随之减少,甚至可减至1。当采用10个正交信号校正因子预处理苹果光谱时,OSC/PLS糖度模型达到最佳性能,最佳模型采纳的因子数为2;采用11个净分析物预处理因子预处理光谱时,NAP/PLS糖度模型达到最佳性能,最佳模型采纳的因子数为1。从总体上评价,最佳OSC/PLS糖度模型和最佳NAP/PLS糖度模型的性能都明显优于原始光谱的最佳偏最小二乘模型。这些结果表明,正交信号校正法和净分析物预处理法都能在保证精度的同时有效地简化苹果糖度预测模型。     

基于区间组合移动窗口法筛选近红外光谱信息

近红外光谱信息的筛选方法是近红外光谱技术简化模型和提高预测精度的重要手段,为选择近红外光谱分析中最优的建模光谱波段,提出区间组合移动窗口偏最小二乘法(SCMWPLS)。该方法通过大小连续变化的移动窗口和交叉有效性筛选出连续的建模区间。以葡萄糖溶液的近红外光谱为测试对象,分别采用所提方法以及传统的间隔偏最小二乘法(IPLS)和窗口移动偏最小二乘法(MWPLS)建立近红外光谱预测模型。所提方法与传统的间隔偏最小二乘法和窗口移动偏最小二乘法模型相比,预测均方根误差分别降低了44%和25%。     

双光路测量结合净信号处理方法在葡萄糖浓度定量分析中的应用

在近红外无创血糖测量中,由于人体血糖浓度变化引起的光强变化非常小,光谱易受到测量仪器以及被测对象自身生理变化的影响。背景变化的影响在测量过程中是无法消除的,只能通过有效的方法来降低这种变化带来的影响,其中选择光学性质与待测物相似,通过扣除相似参考物的光谱来修正该影响是目前常用的手段之一。为减小背景变化对光谱信号的干扰,准确地提取葡萄糖的特异性信号,提出了一种双光路测量结合净信号(NAS)处理的参考测量方法,采用双光路双检测器系统同步测量参考物和被测样品的光谱,然后基于参考物的光谱构建噪声空间计算样品光谱中的葡萄糖的净信号,并在纯吸收和强散射介质中开展了葡萄糖的建模实验,最后,结合二维相关光谱(2DCOS)分析技术和偏最小二乘回归(PLSR)模型对该方法的有效性进行了评价。二维相关分析的结果表明,与直接扣除参考背景的方式相比,净信号处理能突出葡萄糖的特异信号;而PLSR模型的预测结果表明,双光路的效果要优于单光路扣除背景光谱,糖在水溶液和20%-Intralipid溶液中的预测均方根误差(RMSEP)分别降低了35.25%和37.95%;而双光路结合净信号处理后,两种溶液的预测精度又分别提高了26.11%和14.84%。这表明,双光路测量和净信号分析相结合的方法能更有效地提取葡萄糖的特征信息,提高模型的预测精度,从而为无创血糖检测的实现提供更多可能。     

近红外光谱和电子鼻技术用于葡萄酒发酵过程中酒精度的定量分析

采用近红外光谱和电子鼻对葡萄酒的酒精发酵过程进行了动态采样检测,通过主成分回归和偏最小二乘回归对酒精度变化进行了监控和预测研究。分别建立了近红外光谱、电子鼻以及二者融合数据对酒精度定量分析的主成分回归和偏最小二乘回归模型。结果表明,近红外光谱数据和电子鼻数据的主成分回归和偏最小二乘回归模型的相关系数(r)均大于0.99,但校正均方根误差(RMSEC)和预测均方根误差(RMSEP)较大。近红外光谱和电子鼻数据融合后,模型质量得到提高,建立的偏最小二乘模型r为0.999 2,RMSEC和RMSEP分别降低为0.206%和0.205%(v/v),定量精度较高。近红外光谱和电子鼻均适用于红酒发酵过程中对酒精度的定量分析,且二者结合应用能提高定量精度。     

O-PLS在近红外无创人体血糖浓度测量基础研究中的应用

近红外漫反射光谱被认为是无创血糖浓度检测的一种有效方法,但由于被测对象是组织特性复杂的人体,传统的黑箱建模方法很难从综合的光学信息中分离出糖的变化,从而限制了测量的准确性。文章研究了集成偏最小二乘建模和正交信号修正预处理方法于一体的回归方法O-PLS对葡萄糖近红外光谱模型的优化和解释,结果表明O-PLS回归方法能在较大程度上消除无关因素的影响,在简化模型的同时提高了模型的可解释性。     

氯化钠近红外光谱检测技术研究

氯化钠(NaCl)近红外光谱分析在生物医学上有着重要的意义。钠离子(Na+)是人体血液中电解质的主要成分,而电解质有助于维持身体的酸碱平衡。采用近红外光谱技术测量氯化钠浓度,在分析钠离子近红外光谱检测机理的基础上,选定波长建立了NaCl浓度线性回归预测模型,同时为了减小温度对水吸收的扰动,使用选定光谱区建立偏最小二乘(PLS)非线性回归模型。结果表明所建立的非线性校正模型决定系数(R2)=99.82%,交叉验证均方误差(RMSECV)=14.5,剩余预测偏差(RPD)=23.7。完全满足日常生化检测精度要求,该技术可以应用于医院实验室钠离子浓度定量分析。     

SPA-LS-SVM检测土壤有机质和速效钾研究

应用可见/短波近红外光谱分析测量土壤有机质和速效钾含量。光谱预处理包括平滑,标准归一化,多元散射校正和平滑结合一阶导数,以消除系统噪声和外部干扰,分别应用偏最小二乘和最小二乘支持向量机方法建立校正模型,模型的输入为基于连续投影算法得到的特征波长。比较显示基于连续投影算法得到的特征波长为输入的最小二乘支持向量机优于偏最小二乘法建模。模型评价指标由相关系数和预测均方误差表示。有机质的相关系数和预测均方误差分别0.860 2和2.98,速效钾为0.730 5和15.78。表明基于连续投影算法可见/短波近红外光谱利用最小二乘支持向量机建模,可以作为一个精确的土壤有机质和速效钾的测定方法。     

偏稳健M回归在人体血糖浓度近红外无创检测中的应用

采用偏稳健M回归方法有效地解决了人体血糖浓度近红外无创检测研究过程中由于样本奇异值影响模型稳健性的问题。该方法源于现有的迭代变权偏最小二乘法,计算快、易于实现,具有M估计的所有性质,且当权函数选择合适时,能降低奇异值的影响,建立具有稳健性的校正模型。采用该方法对近红外光谱实验数据进行了处理,并与传统的偏最小二乘(partialleast squares,PLS)建模方法进行了比较。结果表明,与PLS相比,该方法可建立稳健的校正模型提高预测精度,更适合复杂样品建模,对于人体血糖浓度近红外无创检测的进一步研究具有应用价值。     

黄酒糖度预测的可见-近红外光谱方法研究

提出了用可见近红外光谱结合不同化学计量学方法预测黄酒糖度的新方法。用240个黄酒样本建模,60个样本进行预测。通过对光谱数据进行平滑、变量标准化及一阶导数等预处理,建立并比较了偏最小二乘法,小波变换与偏最小二乘法相结合,主成分分析与人工神经网络相结合以及主成分分析与最小二乘支持向量机相结合四种不同建模方法的预测精度,以相关系数r、预测标准差、偏差等为评判标准,得到黄酒糖度预测的最优模型为最小二乘支持向量机模型。该模型对黄酒糖度预测的相关系数为0.962、预测标准差为0.021、偏差为-0.001,获得了理想的预测精度。结果表明应用可见近红外光谱对黄酒糖度进行预测是可行的,且最小二乘支持向量机模型能得到最优的预测结果。     

基于自编码网络流形学习的毛竹笋不溶性膳食纤维含量红外光谱建模

提出了一种结合自编码网络(AN)流形学习和偏最小二乘(PLS)法的红外光谱建模方法AN-PLS。AN-PLS方法首先用AN算法对红外光谱数据进行非线性降维,再结合PLS建立回归模型。利用该方法建立了毛竹笋中不溶性膳食纤维含量的近红外光谱和中红外光谱回归模型。结果表明,用AN-PLS方法建立的回归模型,比用其他常用光谱数据预处理方法结合PLS及用单独PLS算法建立的模型具有更小的预测均方根误差RMSEP和更高的决定系数R2,因此,AN-PLS具有较优的建模与预测能力,利用近红外光谱和中红外光谱技术结合AN-PLS建模,可实现毛竹笋中不溶性膳食纤维含量的准确测量。     

Quantitative Analysis of Near-Infrared Spectroscopy by Combined Stationary Wavelet Transform–Support Vector Machine

Data preprocessing and multivariate regression methods are two key factors influencing the model prediction ability of near-infrared (NIR) spectroscopy. The present paper evaluated the application of the combined stationary wavelet transform–support vector machine method for developing juice NIR models. The performance of this method has been compared with other methods, such as stand normal variate–partial least squares, stationary wavelet transform–partial least squares, and stand normal variate–stationary wavelet transform methods. The result showed that compared with other methods, the stationary wavelet transform–support vector machine method can provide good quantitative analysis on saccharose concentration in juice.     

Optimized prediction of sugar content in ‘Snow’ pear using near-infrared diffuse reflectance spectroscopy combined with chemometrics

In near-infrared spectroscopy applications, the original spectra often contain redundant information, which will seriously affect the performance of chemometric models. Therefore, preprocessing, effective wavelengths selection, and appropriate regression models are essential. The objective of this study was to optimize the nondestructive determination multivariate calibration model of sugar content in ‘Snow’ pears, using near-infrared diffuse reflectance spectroscopy combined with chemometrics. All data (sugar content reference values and spectra data) from three measuring positions (P1, P2, and P3, marked around the pear’s equator at angular distances of approximately 120°) were divided into four grouped datasets, namely Set-1 (P1), Set-2 (P2), Set-3 (P3), and Set-4 (average of the three measuring positions). All subsequent optimized processes were performed based on each grouped dataset. First, different preprocessing methods were tested and an optimal method was determined. Then, synergy interval partial least squares and synergy interval partial least squares-competitive adaptive reweighted sampling were applied to select effective regions and effective wavelengths from all wavelengths, respectively, and partial least squares regression models were established and analyzed. In addition, support vector regression models were also established for comparative study. After comprehensive analysis of prediction accuracy and model complexity, the partial least squares regression model based on the 16 selected effective wavelengths for Set-4 was optimal, with the correlation coefficient for prediction, root-mean-square error of prediction, and residual predictive deviation of 0.9701, 0.2311, and 4.12, respectively. The results indicated that with these optimized processes, the multivariate calibration model of sugar content in ‘Snow’ pears was effectively optimized for each dataset. In addition, it is concluded that partial least squares regression was superior to support vector regression in this study, although some other researches had found different results in related fields.     

Characterization of a wavelength selection method using near-infrared spectroscopy and partial least squares with false nearest neighbors and its application in the detection of the chemical oxygen demand of waste liquid

Abstract

The spectral wavelength selection method is important in near-infrared spectroscopy. Eliminating redundant information and extracting useful information can improve the prediction accuracy and modeling efficiency of the quantitative analysis model for spectral analysis to obtain a near-infrared calibration model with strong predictability and good robustness. This paper proposes a wavelength selection method for near-infrared spectroscopy by combining the partial least squares and false nearest neighbor methods. In this method, the correlation between the characteristic wavelength variables and the measured index is assessed by means of a similarity-based distance measure of the characteristic wavelength variable, and the characteristic wavelength is selected according to the order of the correlation. The method was used to select characteristic wavelengths from the near-infrared spectrum of waste liquid to establish a prediction model for the chemical oxygen demand. Compared with the full-spectrum partial least squares and interval partial least squares based models, the number of characteristic wavelength variables is reduced from 1557 to 176, and the prediction accuracy of the model is improved. This method both simplifies the model and achieves higher prediction accuracy. Therefore, this study provides a novel solution for wavelength selection for multivariate calibration in near-infrared spectroscopy.     

堆叠监督自动编码器的近红外光谱建模

近红外光谱中包含了物质中有机分子含氢基团的特征信息,具有维度高、冗余大等特点。传统的基于浅层校正模型,比如主成分回归、偏最小二乘回归、人工神经网络、支持向量回归等,无法提取近红外光谱数据深层的信息。提出一种基于堆叠监督自动编码器的近红外光谱建模方法,不仅可以拟合光谱数据与理化值之间复杂的非线性关系,还可以提取数据深层的特征信息。首先通过对比不同的光谱预处理对模型预测结果的影响,选择最优的预处理方法,然后再使用相关系数法提取特征波段。将处理好的近红外光谱数据作为堆叠监督自动编码器的输入信号,利用理化值对多个监督自动编码器进行有监督的预训练;将多个经过预训练的监督自动编码器进行堆叠,得到堆叠监督自动编码器;将预训练的参数作为堆叠监督自动编码器的初始化参数,然后再利用理化值对堆叠监督自动编码器进行有监督的微调,最后得到模型的最优参数。分别利用玉米含水量和黄酒总酸含量等近红外数据集进行验证,建立了偏最小二乘回归预测模型、人工神经网络预测模型、堆叠自动编码器预测模型和堆叠监督自动编码器预测模型,验证了堆叠监督自动编码器建模的可行性;以预测均方根误差和预测相对分析误差两个指标对比分析了偏最小二乘回归、反向传播人工神经网络、堆叠自动编码器及堆叠监督自动编码器四种建模方法的评价指标。分析结果表明,采用该方法建立的模型,模型预测效果更好,玉米含水量数据集的两个评价指标达到了0.060 4和4.313;黄酒总酸含量数据的两个评价指标达到了0.120和4.227,均优于另外三种方法。     

近红外光谱快速检测丙氨酸氨基转移酶

在探讨近红外光谱快速检测丙氨酸氨基转移酶的可行性过程中,首先对不同厚度(0.5,1,2和4mm)血液样品的近红外透射光谱进行了分析。发现全血样品0.5 mm厚时的近红外透射光谱更适合于进行光谱分析。进而采集了176个全血样品0.5 mm厚时的近红外光谱。对采集的光谱进行多元散射校正、二阶微分法光谱预处理后,采用逐步多元线性回归和偏最小二乘回归方法建立定量分析模型,预测了全血丙氨酸氨基转移酶的含量。结果表明:利用近红外光谱法测定丙氨酸氨基转移酶时,采用偏最小二乘回归方法建立的定标模型预测效果最好,定标相关系数、定标标准差和预测标准差的值分别为:0.98,2.42和7.22。     

利用多光程光谱法检测血液多种成分含量的研究

为研究利用血清多光程光谱信息检测血液成分含量的可行性,通过设计自动微位移测量装置实现了对200份血清样品0～4.0 mm(间隔0.2 mm)近红外多光程光谱的测量。利用多光程光谱的非线性特性,使用血清多光程光谱共同参与建模的方法,对血清样品的葡萄糖(GLU)、总胆固醇(TC)、总蛋白(TP)和白蛋白(ALB)四种成分含量进行定量分析。采用偏最小二乘法(PLS)对160个血清样本的多光程近红外吸收谱与全自动生化分析仪得到的生化分析结果建立校正模型,并对预测集40个样本进行预测。４种血液成分预测值与生化分析结果的相关系数(r)分别为0.932 0,0.971 2,0.946 2和0.948 3。研究结果证明多光程光谱法建模方法用于血液成分含量分析的可行性,为利用多光程光谱法实现方便、快捷的检测血液以及其他液体成分含量奠定了基础。     

Application of transcutaneous diffuse reflectance spectroscopy in the measurement of blood glucose concentration

In this paper, the propagation characteristics of near-infrared (NIR) light in the palm tissue are analyzed,and the principle and feasibility of using transcutaneous diffuse reflectance spectroscopy for non-invasive blood glucose detection are presented. An optical probe suitable for measuring the diffuse reflectance spectrum of human palm and a non-invasive blood glucose detection system using NIR spectroscopy are designed. Based on this system, oral glucose tolerance tests are performed to measure the blood glucose concentrations of two young healthy volunteers. The partial least square calibration model is then constructed by all individual experimental data. The final result shows that correlation coefficients of the two experiments between the predicted blood glucose concentrations and the reference blood glucose concentrations are 0.9870 and 0.9854, respectively. The root mean square errors of prediction of full cross validation are 0.54 and 0.52 mmol/1, respectively.     

近红外光谱谱线特性对物质浓度分析误差影响的研究

为解决近红外光谱法分析物质浓度过程中缺乏可测度分析而导致测量过程存在一定盲目性问题,研究在已知测量条件、样品种类、被测组分以及建模分析方法的条件下,利用近红外光谱谱线特性作为参数,在大量样品近红外光谱采集和标准法测得浓度数据等工作前,对被测物质浓度的分析误差做大致估算。经过大量尝试和试验提出等效信噪比（ESNR）和谱线重叠系数（OC）两个重要参数,其中ESNR反映待测组分吸光度占总吸光度的比重,而OC则反映待测组分近红外光谱曲线间的重叠程度。通过理论仿真得到光谱分析中用经典的偏最小二乘回归建立定量分析模型时谱线特性与物质浓度分析误差的关系,分别计算ESNR和OC与被测组分浓度分析误差（RMSE）的关系,并且研究两个谱线参数的独立性。利用理论分析得到结果对浓度为8%~12%乙醇水溶液进行可测度分析,并与近红外光谱法分析的实际结果进行比较。研究通过理论仿真得到使用光谱分析中经典的偏最小二乘回归建立定量分析模型时谱线特性与物质浓度分析误差的关系,其中ESNR与RMSE成反比关系,而OC与被测组分分析误差成非线性的单调关系,并且验证了ESNR和OC两个参数的独立性。通过理论计算和乙醇水溶液近红外光谱检测实验对等效信噪比和谱线重叠系数与光谱分析浓度误差的定量关系进行讨论,通过理论分析得到的乙醇浓度RMSE预估值为0.30%,近红外光谱分析实际RMSE为0.32%,相对误差6.67%,二者结果相符。实现了在测量条件、样品种类、被测组分以及建模分析方法已知的条件下基于近红外光谱分析的待测组分含量理论误差的定量计算和实验验证。该研究明确了对近红外光谱法分析物质浓度有明确定量关系的两个谱线参数,给出了使用光谱分析中经典的偏最小二乘回归建立定量分析模型时的分析误差经验曲线,以及利用曲线进行近红外光谱法待测组分浓度可测度分析方法。结果表明所提出的ESNR和OC两个谱线特性参数的有效性,以及分析误差预估方法的有效性。为近红外光谱法待测组分浓度定量分析提供了有效、快捷的预估方法,完善了近红外光谱法成分含量可测度分析理论,对近红外光谱法物质浓度定量分析研究具有一定指导意义。     

Selection of near infrared spectral information based on section combination moving window partial least squares北大核心CSCD

Selection of near infrared spectral information is research focus on the application of NIR, which enable to simplify the model and improve accuracy of prediction. Aiming at selecting optimal modeling spectral width of near infrared spectroscopy, section combination moving window partial least squares (SCMWPLS) is proposed in this paper. This method selects continuous modeling screened interval.by continuously varying size of moving window and cross validation. Taking Glucose solution near-infrared spectroscopy as test specimen, near infrared prediction models are established respectively by proposed method, and traditional interval partial least squares (IPLS) and moving window partial least squares (MWPLS). Comparing proposed method with two traditional methods, squares prediction RMSE is decreased by 44% and 25% respectively. © 2017, Editorial Board, Journal of Applied Optics. All right reserved.