利用近红外光谱和偏最小二乘回归法预测脂肪酸组成

采集了30种植物油样品在10000~55 00 cm-1范围内的近红外透射光谱,将所有样品作为校正集,随机抽取10种样品作为预测集,以气相色谱方法测得植物油中主要成分油酸、亚油酸、棕榈酸、硬脂酸的含量为参考值,应用偏最小二乘回归法建立了基于近红外光谱的测定植物油主要成分含量的校正模型。四种成分校正模型的交叉验证误差均方根为0.281 1%~1.496 4%,预测误差均方根为1.080 8%~18.063 0%,校正集的预测值与实测值的相关系数均大于0.99,预测集中除了棕榈酸的预测值与实测值的相关系数为0.817 9,其余均大于0.9。     

近红外光谱结合不同偏最小二乘法快速检测镇江香醋的浑浊度

利用近红外光谱(4000cm-1～10000cm-1)结合化学计量学方法快速检测了镇江香醋中的浑浊度。首先,用近红外光谱仪采集香醋样本的近红外光谱数据以及用离心法测定样本的浑浊度值;然后,采用间隔偏最小二乘法(iPLS)、反向区间偏最小二乘法(biPLS)、联合间隔偏最小二乘算法(siPLS)优选光谱特征区间;最后,采用全光谱(4000cm-1～10000cm-1)偏最小二乘法(PLS)对优选出来的区间建立香醋浑浊度近红外光谱模型。结果表明,采用siPLS将全光谱均匀划分30个子区间,选择4个子区间[4 10 18 27]联合时,建立的模型预测效果最佳,其RMSECV和RMSEP分别为0.173和0.208,校正集和预测集相关系数分别为0.9337和0.9004。因此,利用近红外光谱技术快速检测香醋中的浑浊度是可行的。     

基于近红外光谱与组合间隔偏最小二乘法的稻米镉含量快速检测

采用近红外光谱(near infrared spectroscopy,NIRS)结合组合间隔偏最小二乘法(synergy interval partial least squares,siPLS)建立稻米镉含量快速检测的方法。收集并分析72个稻米样品的NIRS谱图。对光谱前处理方法进行优化,确定平滑、多元散射校正与均值中心化处理为最优方法。采用siPLS法筛选特征波数,建立稻米镉含量的定量模型。稻米镉siPLS模型交叉验证均方根(RMSECV)与预测均方根(RMSEP)值分别为0.247与0.261,训练集相关系数(Rc)与预测集相关系数(Rp)值分别为0.919与0.895。结果表明:运用siPLS法选择特征波长后,不但可以降低模型的复杂度,同时还能够提高预测精度。NIRS作为一种快速、无损与便捷的初筛方法,可用于检测稻米中镉含量是否超标。     

近红外光谱技术结合偏最小二乘判别分析检测三七品质

目的 利用近红外光谱技术实现不同等级三七样品的快速鉴别。方法 采集等级A（20头）、等级B（30头）、等级C（40头）、等级D（60头）四种不同等级三七样品的近红外光谱,构建偏最小二乘判别分析（partial least squares discriminant analysis, PLS-DA）分类器模型鉴别四种等级的三七样品,同时为了减近红外光谱中的冗余波长变量,进一步优化模型的判别结果,利用竞争自适应重加权采样（competitive adaptive reweighted sampling, CARS）算法提取近红外光谱中的特征变量。结果 所构建的PLS-DA分类器模型对等级C和等级D的三七样品,鉴别准确率达到100%,但是对于等级A和等级B的三七样品因为存在误判,鉴别准确率仅为0%和20%。经过CARS算法提取近红外光谱特征变量后,光谱变量数大幅减少,从1557个变量下降到78个变量。以优选后的特征变量构建的CARS-PLS-DA分类器模型更加简化,对四种等级三七样品的预测均方根误差均明显下降,说明模型的预测分类变量更接近真实的分类变量,鉴别结果更加准确。同时,对四种等级三七样品的鉴别准确率显著上升,其中对于等级C和等级D的鉴别准确率为100%,对于等级B的鉴别准确率从20%提升到100%,等级A鉴别准确率从0%提升到75%。结论 所构建的CARS-PLS-DA分类器模型对四种等级的三七样品具有更好的鉴别效果,可以实现不同等级三七的品质鉴定。     

基于偏最小二乘(PLS)法白酒中乙醇含量的近红外检测

将近红外光谱与偏最小二乘法相结合,对白酒中乙醇含量进行快速准确检测。研究了标准溶液的近红外吸收光谱和一阶导数光谱,采用偏最小二乘法建立校正模型,选择了最佳主成分数,并对实际酒样中乙醇进行预测,得到了比较满意的结果。     

基于近红外光谱的柠檬酸发酵液化清液概率偏最小二乘法监控

柠檬酸发酵液化清液生产过程监控对整个柠檬酸生产至关重要,近红外光谱能够通过不同波长下分子的振动多方面地反映过程的运行状况,包含了大量的过程信息。但是,现有方法往往是建立近红外光谱与总糖总氮等质量变量的回归模型,通过判断质量变量是否超过阈值实现对过程运行状态的事后报警,忽略了近红外光谱内部的很多有用信息,监控效果较差。该文充分利用和分析近红外光谱的统计特性,提出一种基于近红外光谱生产过程的统计监控方法,首先建立近红外光谱和总糖总氮的概率偏最小二乘模型(probability partial least squares,PPLS),然后基于模型对不同的信息设计监控指标,能够充分利用近红外不同波长上的信息,实现故障的事前预警。结果表明,采用该方法得到漏报率为9.68%,错报率为25.81%,可以有效地对柠檬酸发酵液化清液生产过程进行监控。     

近红外光谱结合不同偏最小二乘法无损检测食醋中总酸含量

探讨了快速、无损检测食醋中总酸含量的建模方法,利用近红外光谱法分别结合间隔偏最小二乘法(iPLS)、反向区间偏最小二乘法(BiPLS)、联合间隔偏最小二乘算法(SiPLS)进行建模,对各算法在不同划分区间数及区间选择时对建立模型的影响进行比较.结果表明:BiPLS、SiPLS(2,3,4区间联合)建模效果较好于iPLS所建立的模型,其中BiPLS在选择43个子区间,5个子区间联合(3,4,6,7,16)最佳,其RMSECV和RMSEP分别为0.2876和0.2726,校正集和预测集相关系数分别为0.9343和0.938;SiPLS在选择3个区间联合,49个区间数(3、5、7区间联合)最佳,其RMSECV和RMSEP分别为0.2607和0.2802,校正集和预测集相关系数分别为0.9463和0.9371;iPLS在选择22个子区间,第三个子区间,主因子数为4时最佳,其RMSECV和RMSEP分别为0.2998和0.2977,校正集和预测集相关系数分别为0.928和0.9213.不同偏最小二乘算法所选取区域大多集中于5500～6000 cm-1范围内,证明该波数范围应该是总酸的相应特征区间.     

近红外光谱结合偏最小二乘法快速测定糖果中水分含量

采用近红外光谱(NIR)结合偏最小二乘法(PLS)建立了一种糖果中水分含量快速准确的测定方法。在12500~3600 cm^-1光谱范围内采集116批糖果的近红外漫反射光谱,并用减压干燥失重法测定其水分含量。通过比较不同参数对建模的影响,发现用多元散射校正法进行预处理,在11682.2~9826.1、8939.0~6267.9、5378.8~4487.8 cm^-1光谱范围内,主成分数为15时,应用PLS方法建立的糖果水分的定量分析模型效果最佳。所建立模型的相关系数为0.9716,校正均方根误差和验证均方根误差分别为0.97%和1.03%。该方法结果准确可靠、操作简便,可用于糖果中水分含量的快速检测。     

近红外光谱结合区间偏最小二乘法应用于花生油酸价的测定

采用近红外光谱法结合不同区间偏最小二乘波长筛选法建立花生油酸价的定量分析模型。采用酸碱滴定法测定花生油样本的酸价同时采集近红外光谱数据;采用区间偏最小二乘法(iPLS)、向后区间偏最小二乘法(BiPLS)、移动窗口偏最小二乘法(mwPLS)优选光谱特征区间;采用偏最小二乘法(PLS)对优选出来的谱段建立酸价的定量模型。结果表明,采用mwPLS选择的谱段建立的模型预测效果最佳,RMSECV和RMSEP分别为0.247 76和0.131 5,校正相关系数和预测相关系数分别为0.993 2和0.996 9。因此,近红外光谱结合移动窗口偏最小二乘法可以快速准确测定花生油的酸价。     

近红外光谱法快速鉴别转基因大豆模型优化

目的应用近红外光谱(NIR)结合偏最小二乘判别分析(PLS-DA)建立转基因大豆的快速鉴别模型,并选择最优模型。方法主成分分析(PCA)用于从光谱数据中提取相关特征并剔除异常样品。在试验中,94份样品用于构建模型,41份样品用作验证评估模型的效果。分别讨论样品形态(整粒和粉末)、波长范围和光谱预处理方法对所建模型判别正确率的影响。结果粉末状大豆样品建模的效果好于整粒大豆样品。其中判定效果最好的模型,整粒大豆在9 403～5 438 cm~(-1)范围内,采用二阶导数(2nd)处理光谱,模型的校正集和验证集的判定正确率均为100.00%;粉末状大豆在7 505～4 597 cm~(-1)范围内,采用矢量归一化+一阶导数(SNV+1st)处理光谱,模型的校正集和验证集的判定正确率也均为100.00%。结论通过选择样品形态、波长范围和光谱预处理方法可以优化鉴别模型,提高近红外判别模型的鉴别正确率。     

Measuring fig quality using near-infrared spectroscopy

Sorting of dried figs prior to inspection is labor-intensive and somewhat complex. We examined the potential of using near-infrared spectroscopy (NIRS) to automate sorting of dried figs. Calimyrna and Adriatic types were inspected by hand using established criteria. For both varieties, approximately 100 passable figs and 100 figs each for the infested, rotten, sour, and dirty defect categories were examined using NIRS and partial least-squares regression (PLS). Correct classifications for these varieties ranged from 83 to 100%. About twenty PLS factors were used to make the predictions. These results indicate that the use of NIRS to help automate inspection for dried fig processing is feasible. However, the large number of wavelengths needed for prediction, as indicated by PLS beta coefficients, indicates that implementing NIRS in fig sorting may require an instrument capable of reading numerous wavelengths rather than a more economical filter-based instrument.     

Kernel partial least squares model for pectin content in peach using near-infrared spectroscopy

Pectin is a class of complex galacturonic acid-rich polysaccharides that are related to the texture of fruit and vegetables. The objective of this study was to develop, using near-infrared (NIR) spectroscopy, the best model for the determination of pectin content in peach fruit. A total of 100 samples divided into lossy and lossless samples were used to collect NIR raw spectra in the range of 1000–2500 nm. NIR absorption spectra were then obtained after pre-processing. Finally, four methods were used to establish lossy and lossless spectral models. The 10-fold cross-validation coefficient of determination R² of the lossy model was between 0.364 and 0.628, whereas that of the lossless model was between 0.187 and 0.288, indicating that the lossy model was better than the lossless model. Among all samples, the kernel partial least squares (KPLS) lossy model was better, with coefficient of determination R² = 0.628, root mean square error (RMSE) = 0.069 and mean absolute error (MAE) = 0.061. This is the first study to evaluate the prediction of peach pectin content using NIR spectroscopy, and the model can be used for rough screening.     

各品种不同成熟度无花果质构特性分析

利用质构仪穿刺技术对无花果果实的3个不同部位分别进行了比较分析,同时采用配对T检验和相关性分析对不同成熟度果实品质差异进行了研究。结果表明:果实的近果柄端比果实中部和近果目端都比较硬,不同成熟度的果实破裂深度相差不大,随着成熟度的增加,果皮韧性和果肉硬度减小,其中金傲芬的果皮韧性、果实硬度和黏性在全熟与八成熟果实对比中变化最大,适宜鲜食;青皮、绿早在八成熟与成熟时的果皮破裂强度都很小,八成熟的果实适合长途运输;布兰瑞克的成熟果实黏性最大,可能适宜加工成无花果泥和果酱。各品种的不同成熟期的质构特征可做为鲜食及后期加工方式的选择依据。     

用偏最小二乘紫外光谱法同时测定维生素B混合物

用偏最小二乘紫外光谱法同时测定了B族维生素4组分混合物，采用相关系数－标准偏差法从4组分混合物（波长范围330nm－210.5nm）的紫外光谱中选出11个波长点，根据预测集残余误差选择8个因子，应用偏最小二乘法成功地实现了对VB1，VB2，VB6和NA（nicotinamide）的同时测定，相关系数分别为0.99993,0.99988,0.99998和0.99984，标准偏差分别为0.0234,0.0162,0.0205和0.0986。     

Prediction of maturity and sensory attributes of Cheddar cheese using near-infrared spectroscopy

Twenty-four experimental Cheddar cheeses were produced using 5 renneting enzymes and stored at 4 °C for up to 9 months. At 2, 4, 6 and 9 months, cheeses were analysed for sensory attributes (“crumbly”, “fragmentability”, “firmness”, “rubbery”, “gritty/grainy”, “moist”, “chewy”, “mouthcoating”, “greasy/oily”, “melting” and “massforming”) by a trained panel of 10 assessors. Near-infrared (750–2498 nm) reflectance spectra were recorded contemporaneously. Predictive models for the sensory attributes and age (months) were developed by partial least-squares (PLS) regression; raw, derivatised and scatter-corrected spectral data were investigated. As a general rule, the most accurate models were produced by spectral data in the range 750–1098 nm after a 2nd derivatisation step. Age was predicted with a root mean square error of cross-validation (RMSECV) equal to 0.61; sensory attributes successfully modelled and their respective RMSECV values were “crumbly” (2.3), “rubbery” (3.4), “chewy” (4.0), “mouthcoating” (5.0) and “massforming” (4.1). These models are sufficiently accurate to be industrially useful.     

Variable selection for partial least squares analysis of soluble solids content in watermelon using near-infrared diffuse transmission technique

This work is focused on the variable selection in building the partial least squares (PLS) regression model of soluble solids content (SSC) that is used to evaluate quality grading of watermelon. The spectra were obtained by the near infrared (NIR) spectrometer with the device designed for on-line quality grading of watermelon and the spectra of 680–950 nm were adopted to analysis. The variable selection was based on Monte-Carlo uninformative variable elimination (MC-UVE) and genetic algorithm (GA). In comparison of the performances of the full-spectra (680–950 nm) PLS regression model and the feature wavelengths PLS regression model showed that the MC-UVE–GA–PLS model with baseline offset correction combined multiplicative scatter correction (MSC) pretreatment was much better and 14 variables in total were selected. The correlation coefficients between the predicted and actual SSC were 0.885 and 0.845, the root mean square errors were 0.562 °Brix and 0.574 °Brix for calibration and prediction set, respectively. This work can make a great contribution to the research of on-line quality grading for watermelon nondestructively.     

红外及近红外光谱技术在食品分析检验中的应用进展

红外及近红外波谱技术,是利用物质对红外光区的电磁辐射的选择性吸收来进行结构分析及定性和定量分析,具有快速,方便,样品用量少,不损坏样品等优点。     

我国蓝莓食品加工现状分析

基于可见近红外光谱技术,采用深度学习中的堆叠监督自编码器（stacked supervised autoencoders,SSAE）对蓝莓果渣的花青素含量进行了建模。首先对光谱数据进行预处理和特征筛选处理,以预设SSAE模型的预测集均方根误差（RMSEP）最低为标准,选择出178个特征波长;以选择出的特征波长处的吸光值作为SSAE模型的输入,以蓝莓果渣中的花青素含量为输出,讨论SSAE模型激活参数、节点数、训练次数和学习率,得到SSAE最优参数,即激活函数rule、结构178-60-5-1、训练次数70、学习率0.01。选取训练集均方根误差（RMSEC）、预测集均方根误差（RMSEP）、预测集相关系数（Rp）为评价标准,获得所建立模型的RMSEC、RMSEP、Rp分别为1.0500、0.3835、0.9042。最后通过与经典回归预测模型极限学习机（extreme learning machine,ELM）、最小二乘支持向量机回归（least squares support vector regression,LSSVR）和偏最小二乘回归（partial least squares regression,PLSR）算法进行对比,发现本研究所建SSAE模型的预测精度更高,表明SSAE模型与可见近红外光谱结合能有效预测蓝莓果渣中的花青素含量。     

近红外光谱、中红外光谱、拉曼光谱无损检测技术在食用油脂分析中的研究进展

近红外光谱(NIRS)、中红外光谱(MIRS)及拉曼光谱(Raman)技术能对样品进行无损分析,近年来在油脂分析领域表现出极大的应用潜力。本文综述了NIRS、MIRS及Raman技术在油脂分析中的研究进展,具体介绍了油脂脂肪酸组成分析、食用油的掺假鉴别、游离脂肪酸分析、脂肪酸顺反结构和不饱和度分析等方面的应用,并对NIRS、MIRS及Raman技术在油脂分析中的应用前景做了进一步展望。