近红外光谱结合极限学习机的榛子蛋白质含量检测

以毛榛与平榛作为研究对象,将去壳后的毛榛与平榛分别制成90组与60组试验样本,获取波长范围为900 nm～1700 nm的原始光,同时,通过凯氏定氮法测得其中蛋白质含量真实值,并通过对比研究一阶(1-der)导数、二阶(2-der)导数、多元散射校正(multiplicative scatter correction,...     

基于近红外光谱的大米蛋白质含量快速检测

针对国标化学检测方法耗时耗力、成本昂贵的问题,分析了近红外光谱（NIRS）结合化学计量学进行大米蛋白质含量检测的可行性。基于变量选择、特征提取和非线性建模的策略,将反向区间偏最小二乘（BiPLS）与主成分分析（PCA）和支持向量机（SVM）相结合,构建了BiPLS-PCA-SVM模型,用于提高蛋白质回归模型的性能。在BiPLS-PCA-SVM模型中,将蒙特卡罗交叉验证与预测残差平方和相结合进行最佳主成分个数的选择,通过遗传模拟退火算法对模型参数进行优化。为了评估BiPLS-PCA-SVM模型的性能,建立了Full-PLS、BiPLS和BiPLS-SVM 3种模型,并系统分析了上述模型的预测精度和鲁棒性。BiPLS-PCA-SVM模型在预测蛋白质含量方面显示的性能高于其他模型,使用最佳主成分和优化后的SVM参数建立的模型具有更高的鲁棒性和准确性。对于BiPLS-PCA-SVM模型,验证集的决定系数、均方根误差和剩余预测偏差分别为0.928 9、0.196 7%和4.024 6。结果表明,NIRS与BIPLS-PCA-SVM模型相结合,可作为替代策略实现大米中蛋白质含量的快速检测。     

近红外光谱技术快速鉴别查哈阳大米

查哈阳大米为黑龙江特色品牌大米,为保护查哈阳大米品牌,实现查哈阳大米原产地快速真伪鉴别。通过构建不同地域和品种水稻试验田,收获期内对试验田大米的近红外光谱进行全波长扫描,筛选与产地相关特征波段。在特征波段处对来自查哈阳地区及非查哈阳地区的233份大米进行定性分析及定量分析。结果表明,与产地因素主要相关的波长为5 136~5 501 cm~(-1),利用此波段采用因子化法建立定性分析模型对查哈阳大米正确鉴别率为100%,采用偏最小二乘法建立的定量分析模型对查哈阳大米正确判别率为95.83%。     

基于近红外光谱技术快速检测稻谷水分含量

目的:建立一种无损、快速高效的稻谷水分含量检测方法。方法:研究收集了不同年份的稻谷样品161份,运用近红外光谱结合化学计量学方法,通过剔除异常光谱和光谱预处理,采用偏最小二乘法建立稻谷水分含量预测模型。结果:采用主成分分析结合马氏距离的方法剔除异常光谱样品15个,最佳的光谱预处理方式为消除常数偏移量。训练集建立的预测模型(R_CAL²)为0.9943,模型标准偏差(RMSEC)为0.21%,模型交叉验证决定系数(R_CV²)为0.9936,模型交叉验证标准偏差(RMSECV)为0.32%,表明预测模型交叉验证预测样品水分含量准确度高。用验证集样品检验预测模型,模型验证集验证决定系数R 2 VA L为0.9801,模型验证集验证标准偏差(RMSEP)值为0.36%,相对分析误差(RPD)值为7.14,表明预测模型对未知样品的预测准确度高。验证集样品实测值与预测值均值方程T检验结果P值(双侧)为0.879,验证集样品实测值与预测值之间差异不显著,表明预测模型的预测结果可信度高,验证集样品预测值与实测值的误差在±1%,且90%以上的验证集样品其预测值与实测值的误差都在±0.5%以内。结论:建立的稻谷水分预测模型可以实现收储稻谷的无损、快速、准确检测。     

基于变量优选的苹果糖分含量近红外光谱检测

目的：在900~1 700 nm的波长范围内采集苹果的近红外光谱数据,结合化学计量学方法对糖分含量进行无损检测。方法：先对光谱数据依次进行基线校正、散射校正、平滑和尺度缩放,以交叉验证的均方根误差最小选出最佳的预处理方法。采用连续投影算法（SPA）、竞争性自适应重加权算法（CARS）分别选取7,52个特征变量。分别以连续投影算法选取的特征变量、竞争性自适应重加权算法选取的特征变量、两种方法选出的特征变量的组合作为输入自变量,建立线性偏最小二乘回归法模型和非线性的极限学习机模型。结果：组合的特征变量建模效果优于单一方法选出的特征变量的建模效果,非线性模型优于线性模型。结论：采用组合的特征变量,建立极限学习机模型,预测效果最优,训练集的均方根误差为0.710 1,拟合优度为0.883 8,测试集的均方根误差为0.637 5,拟合优度为0.894 5。     

基于近红外光谱的大米加工精度等级快速判定

将30个粳稻3、6个籼稻品种糙米碾削20～95 s,得到不同糙出白率的白米。分别采用称重法和染色法测出每个样本的糙出白率和加工精度等级,用FT-NIR近红外光谱仪采集每个样本的近红外光谱,以偏最小二乘法建立稻谷糙出白率的定标模型并对模型进行验证,得到最小预测误差(RMSEP)为0.032 4%,相应的决定系数(R2)为0.999 5;对稻米加工精度等级与糙出白率关系进行统计分析,发现随着稻米加工精度等级的提高,糙出白率呈对数关系降低。基于预测得到的糙出白率,建立稻米加工精度等级判定模型,结果显示,对数模型的预测精度为82.5%,采用马氏距离判别法建立的模型预测精度可达98.33%。     

短波近红外光谱快速无损检测生鲜鸡肉胆固醇含量

采用短波近红外光谱对生鲜鸡肉中的胆固醇含量进行检测,使用便携式近红外光谱仪在近红外光谱短波区域采集236份生鲜鸡肉的光谱信息,采用化学计量学法建立鸡肉胆固醇的偏最小二乘法定量预测模型。结果表明:最佳光谱预处理方法为标准化和基线校正,并通过剔除两次异常值对模型进行校正,所建定标模型的校正集相关系数Rc=0.801 1,校正标准差(s_(EC))=6.699 8,验证集相关系数R_p=0.803 4,预测标准差(s_(EP))=7.529 6,主因子数MF=4,s_(EP)/s_(EC)=1.12,说明模型可靠性、稳健性和预测效果较好。     

傅里叶变换近红外光谱快速检测方便面油脂含量

利用傅里叶变换近红外光谱技术测定方便面油脂含量。光谱通过多元散射校正算法(MSC)处理后,采用偏最小二乘法(PLS)建立模型,用定标集预测误差(SEC)、验证集预测误差(SEP)及其相关系数(Rc,Rv)评判模型好坏。结果发现,该法定标集预测误差和验证集预测误差分别为0.4456和0.4447,定标集和验证集相关系数分别为0.9798、0.9755;研究表明,利用近红外光谱技术能快速、准确测定油炸方便面油脂含量。     

基于近红外光谱技术建立棉籽蛋白质含量快速无损测定方法

棉籽是棉花产业的重要副产品,含有丰富的棉籽蛋白,是牲畜蛋白质的重要来源,提高棉籽蛋白质的利用率对缓解我国蛋白资源短缺有着重要意义。快速无损环保测定棉籽蛋白质含量方法的建立,对于合理有效的利用棉籽资源和开展棉籽营养品质遗传改良等方面的研究具有关键作用。本研究通过凯氏定氮法对187份棉花品种进行棉籽蛋白质含量化学值测定,利用近红外光谱分析仪分别采集三种不同棉籽形态毛籽、光籽和种仁的光谱特征信息,经过不同数学算法处理和散射处理等方法结合改良偏最小二乘法 (Modified partial least squares, MPLS), 建立了三种不同棉籽形态蛋白质含量快速无损检测的近红外光谱模型。构建的毛籽、光籽和种仁三种近红外检测模型的定标相关系数 (Regression squared, RSQ) 分别为0.957、0.971、0.989,并通过差异统计分析比较,验证集的化学值与近红外光谱预测值间不存在显著差异。结果表明三种不同棉籽形态下建立的近红外光谱检测模型都具有良好的准确性,能够针对不同待测样品类型进行准确的蛋白质含量检测。本研究建立了毛籽、光籽以及棉仁三种棉籽蛋白质含量的近红外光谱检测模型,实现了棉花种子蛋白质含量的快速无损低消耗检测,为棉籽综合利用以及棉籽高品质育种提供技术支撑。     

近红外光谱技术检测燕麦中蛋白质含量

以93份燕麦样品为研究对象,比较了不同光谱预处理方法和不同回归方法对定标模型的影响,建立了定量分析燕麦中蛋白质含量的合理模型。结果表明,最佳的预处理方法:光谱散射处理为标准化处理(SNV),数学处理为2441;最佳的回归方法为改进偏最小二乘法(MPLS)。在此条件下模型对验证集的测定值与预测值的决定系数为0.954 3,均方根误差为0.160 7,模型的预测准确性良好。建立的近红外模型对燕麦中蛋白质快速测定有一定参考价值。     

近红外光谱法检测奶粉掺假

目的建立近红外光谱法结合Adulterant Screen算法快速鉴别奶粉中大豆蛋白和尿素掺假的方法。方法采用近红外光谱仪获得奶粉未知样的光谱曲线,再用Adulterant Screen算法以及全数据库奶粉分类模型和既定类型的掺假物模型对曲线主要成分和掺假成分进行分析。结果该方法对一定浓度大豆蛋白和尿素掺假奶粉样可以实现掺假鉴别,大豆蛋白和尿素掺假奶粉样的掺假判别限分别为0.3 g/100g和0.2 g/100g,掺假物正确识别限分别为0.5 g/100g和0.8 g/100g。结论利用近红外光谱法结合Adulterant Screen算法可以快速鉴别奶粉中大豆蛋白和尿素的掺假。     

近红外光谱法检测奶粉中三聚氰胺的方法探讨

目的建立近红外光谱法快速鉴别奶粉中三聚氰胺的掺假。方法应用奶粉的脂肪、蛋白质、水分、乳糖和灰分5个理化指标建立奶粉的近红外鉴别模型。应用Spectrum软件的PLS1算法对样品进行计算,并对待鉴别奶粉进行定量分析。结果本方法能有效识别三聚氰胺质量比在0.05%~0.08%的掺假奶粉,但不能准确识别有何种掺杂物;对0.1%以上的掺假奶粉能正确识别出掺假物是三聚氰胺。对样品的脂肪、蛋白质、水份、乳糖和灰分5项理化指标进行定量分析的实际值与样品在模型中的计算值相对误差在10%以内。结论该方法快速、有效,可适用于奶粉三聚氰胺掺假的快速筛选和掺假鉴别。     

近红外光谱技术在黄酒理化指标快速检测中的应用

采用傅立叶变换近红外透射光谱技术建立黄酒3项理化指标(酒精度、总酸和氨基酸态氮)的快速检测方法.通过比较不同的频率范围和维数,选择最佳光谱预处理方法,建立了3项指标的预测模型.用R2(决定系数)和RMSECV(交叉验证均方差)衡量3个模型的预测精度和稳定性,R2值分别为97.68,95.68、94.13,RMSECV值分别为0.2030,0.1260,0.0232.对10个随机抽取的黄酒样品的预测结果显示,样品的预测值和实测值间的之间没有显著性差异(p＞0.05).研究结果表明,近红外光谱法可用于黄酒酒精度、总酸和氨基酸态氮的快速检测.     

近红外光谱分析法快速测定稻谷常规化学指标

采集150份有代表性的我国南方地区稻谷样品的近红外光谱,用偏最小二回归分析法(PLS),建立了稻谷的水分、直链淀粉、蛋白以及胶稠度的近红外定量分析模型,并对30份预测集样品进行了验证。水分、直链淀粉、蛋白以及胶稠度的校正集模型的决定系数所(R2)分别为0.990 3、0.560 3、0.913 2以及0.678 0,交互验证均方根误差(RMSECV)分别为0.372 8%、1.456 9%、0.305 4%以及5.031 5%;验证集标准预测偏差(RMSEP)分别为0.382 5%、1.465 0%、0.510 0%以及5.052 1%。结果表明,近红外光谱分析法可以满足快速分析的要求。     

Irradiation dose detection of irradiated milk powder using visible and near-infrared spectroscopy and chemometrics

The objective of this study was to examine the possibility of applying visible and near-infrared spectroscopy to the quantitative detection of irradiation dose of irradiated milk powder. A total of 150 samples were used: 100 for the calibration set and 50 for the validation set. The samples were irradiated at 5 different dose levels in the dose range 0 to 6.0 kGy. Six different pretreatment methods were compared. The prediction results of full spectra given by linear and nonlinear calibration methods suggested that Savitzky-Golay smoothing and first derivative were suitable pretreatment methods in this study. Regression coefficient analysis was applied to select effective wavelengths (EW). Less than 10 EW were selected and they were useful for portable detection instrument or sensor development. Partial least squares, extreme learning machine, and least squares support vector machine were used. The best prediction performance was achieved by the EW-extreme learning machine model with first-derivative spectra, and correlation coefficients = 0.97 and root mean square error of prediction = 0.844. This study provided a new approach for the fast detection of irradiation dose of milk powder. The results could be helpful for quality detection and safety monitoring of milk powder.     

Application of near infrared reflectance spectroscopy to protein analysis of grassland herbage samples

Near infrared reflectance spectroscopy (NIRS) was evaluated for the prediction of the protein content in samples of grassland herbage taken at different stages of maturity (flowering to fruiting stage) in ‘Dehesa’ zones of Central-Western Spain. A Technicon Infra Alyzer model 500 scanning monochromator interfaced with a Hewlett-Packard 1000 minicomputer was used for the study. Protein content was predicted with NIRS data treated as log I/R using six or seven wavelengths. Calibrations were evaluated by comparing Kjeldahl analyses with those predicted by NIRS. The prediction of protein was found to be acceptable, the standard error carying between 0.56 and 0.68% in a range of protein content from 6.76 to 13.98%.     

Hot topic: application of support vector machine method in prediction of alfalfa protein fractions by near infrared reflectance spectroscopy

The object of this study was to explore the potential for support vector machine (SVM) to improve the precision of predicting protein fractions by near infrared reflectance spectroscopy (NIRS). Generally, most protein fractions determined in Cornell Net Carbohydrate and Protein System (CNCPS), especially the neutral detergent insoluble protein (NDFCP) and acid detergent insoluble protein (ADFCP), could not be accurately predicted by the commonly used partial least squares (PLS) method. A recently developed chemometric method, SVM, was applied in NIRS prediction of alfalfa protein fractions in this study. Two hundred thirty alfalfa samples were scanned on a near infrared reflectance spectrophotometer, and analyzed for crude protein (CP), true protein precipitated in tungstic acid (TCP), borate-phosphate buffer-insoluble protein (BICP), NDFCP, and ADFCP. These 5 laboratory proteins and the CNCPS protein fractions A, B1, B2, B3, and C were predicted by NIRS using the PLS and SVM methods. According to PLS-NIRS regression, CP, TCP, BICP, A, and B2 obtained the determination coefficient of prediction of 0.96, 0.91, 0.94, 0.94, and 0.93, and the ratios of standard deviation of prediction samples: standard error of prediction samples (RPD) values were 5.07, 3.31, 3.98, 3.96, and 3.91. Neutral detergent insoluble protein, ADFCP (fraction C), B1, and B3 were predicted with of 0.75, 0.83, 0.30, and 0.62, and RPD values of 1.98, 2.42, 1.20, and 1.62; Calibrated by the SVM-NIRS method, values of CP, TCP, BICP, NDFCP, ADFCP(C), A, and B2 achieved 0.99, 0.97, 0.97, 0.90, 0.93, 0.97, and 0.97, respectively. The RPD values of those fractions were 8.68, 8.26, 6.11, 3.08, 3.69, 5.97, and 5.81, respectively. The and RPD values of fractions B1 and B3 were 2.67 and 0.87 (B1) and 2.51 and 0.75 (B3) directly predicted by SVM-NIRS model. In this study, the chemical analysis results of B1 and B3 were also correlated with calculated results from TCP-BICP and NDFCP-ADFCP, which were predicted by SVM-NIRS models. The B1 protein fraction achieved and RPD values of 0.87 and 3.61, whereas values for B2 were 0.75 and 2.00. Data suggested that use of SVM methods in NIRS technology could improve the accuracy of predicting protein fractions. This study showed the potential of increasing the NIRS prediction accuracy to a level of practical use for all protein fractions, except B3.