基于近红外漫反射光谱技术判别柿子品种和贮藏期的研究

目的通过近红外漫反射光谱技术建立了柿子不同品种和贮藏期的快速判别方法。方法实验对贮藏冷库(0±0.5℃)20 d、40 d、60 d的磨盘柿和阳丰甜柿的近红外光谱(400~2500 nm)进行平滑、一阶倒数和标准正常化处理(SNV)处理,采用主成分分析法(PCA)建立判别模型。结果在全波长范围内,不同品种定标模型的正确分类率达到100%;阳丰甜柿不同贮藏期的正确分类率达到97.78%;磨盘柿不同贮藏期的正确分类率达到98.89%。3个预测模型的累积准确率达到96.67%。结论通过近红外漫反射光谱技术,判别不同品种的柿子并预测其贮藏期具有应用价值。     

近红外光谱对贮藏期樱桃品质的定性判别

研究运用近红外光谱技术对贮藏期樱桃进行定性判别。实验的定标模型经过不同预处理和不同光谱波段条件的处理分析,得到在全光谱范围(408.8~2 492.8 nm)内,采用一阶微分结合去离散处理(SNV and D)的方法可以构建最优模型。该模型判别的正确分辨率为100%,预测准确率为90%~96.7%。实验说明,近红外光谱技术对贮藏期樱桃的检测具有适用性。     

采用近红外光谱进行采后苹果品种及货架期定性判别

运用近红外光谱技术,通过不同光谱预处理和不同光谱波段选择,研究苹果品种(嘎啦、乔纳金、金冠、寒富)及货架期(0、14、28 d)的近红外判别模型。结果表明,不同品种苹果定标判别模型最优光谱预处理方法为:在全波长范围(408. 8～2 492. 8 nm)内,采用去散射结合二阶导数光谱预处理,对未知样品判别正确率为85. 00%～95. 00%;苹果货架期较优定标模型在1 108～2 492. 8 nm范围内,光谱预处理方法为标准正常化处理(standard normal variate,SNV)+去散射处理(detrend,D)+一阶导数,预测样品正确率为91. 67%～96. 67%。实验证明,近红外光谱技术对采后苹果品种及货架期检测具有适用性。     

葛粉掺假的近红外漫反射光谱快速检测

研究葛粉中掺假红薯淀粉和马铃薯淀粉的近红外漫反射光谱快速检测方法。采集样品的近红外漫反射光 谱,采用主成分回归和偏最小二乘法建立校正模型,并对比光谱预处理方法和光谱建模区间对模型的影响。结果表 明,采用偏最小二乘法建模,光谱采用标准正态变量变换预处理,光谱区间选择在962～1 389 nm时,模型预测效 果最佳,外部验证预测相关系数（RP 2）达0.994 5,均方根误差2.298 7%,相对分析误差13.56,平均回收率99.89% （n=9,RSD=2.96%）,这表明近红外漫反射技术能对葛粉中掺假红薯淀粉和马铃薯淀粉进行有效检测。     

甜柿货架期的近红外光谱定性判别

通过近红外漫反射光谱技术对甜柿的货架期进行了定性判别研究,运用不同光谱预处理方法和不同波段选择,发现在1 100² 400 nm内,采用二阶导数结合标准正常化和去散射(SNVD)处理的光谱预处理方法最好。判别模型的正确分类率达到97.8%2 400 nm内,采用二阶导数结合标准正常化和去散射(SNVD)处理的光谱预处理方法最好。判别模型的正确分类率达到97.8%¹00%,预测准确率达到88.9%100%,预测准确率达到88.9%¹00%。因此,近红外光谱技术对甜柿货架期的检测具有应用价值。     

线椒货架期的近红外光谱定性判别

利用近红外漫反射光谱技术对线椒的货架期进行定性判别研究。实验以常温货架期1、3、5 d的线椒为研究对象,利用主成分分析法（PCA）建立近红外漫反射定性判别模型,在全光谱范围（400²500 nm）内比较了不同的光谱预处理方法结合不同散射和标准化方法对所建模型的影响。结果表明,采用全光谱下Log（1/R）+None光谱预处理方法建立的模型预测最好,该模型的交互验证相关系数(R_CV)为0.9455,交互验证误差（SECV）为0.1534,其正确分类率达95.56%¹00%,预测准确率达88.89%⁹7.78%,该模型能够准确地区分不同货架期的线椒鲜果。因此,近红外光谱技术为线椒货架期的鉴别提供了一种新方法。      

可见/近红外漫反射光谱无损检测甜柿果实硬度

该研究的目的是建立可见/近红外漫反射光谱无损检测甜柿果实硬度的数学模型,评价可见/近红外漫反射光谱无损检测甜柿果实硬度的应用价值。果实硬度采用果皮脆性、果皮强度和果肉平均硬度作为评价指标。在可见/近红外光谱区域(400～2 500 nm),采用改进偏最小二乘法,对比分析了不同导数处理、不同散射及标准化处理的甜柿果实硬度定标模型。结果表明,对于果皮强度和果皮脆性,采用最小偏二乘法、一阶导数处理和标准多元离散校正处理建立的定标模型预测效果较好,RP2分别为0.858和0.862,SEP分别为0.094和0.157,RPD分别为2.47和2.63。对于果肉平均硬度,采用改进偏最小二乘法、一阶导数处理和标准正常化和去散射处理建立的定标模型预测效果较好,RP2为0.82,SEP为0.063,RPD为2.35。因此,可见/近红外漫反射光谱无损检测技术可用于甜柿果实硬度的无损检测。     

基于CWT和GRNN的可见-近红外漫反射光谱检测樱桃糖度的研究

联合使用连续小波变换(continuous wavelet transform,CWT)和广义回归神经网络(generalized regression neural networks,GRNN)建立用于测定樱桃中糖含量的CWT-GRNN 预测校正模型。利用CWT 提取樱桃样本数据中反映含糖量的关键光谱特征,在CWT 域中选择3 个具有代表性的尺度,并在每个尺度下根据樱桃样本的可见- 近红外光谱的特征将其划分为4 个特征区间,从而构造12 个特征输入到GRNN,GRNN 的光滑因子取为0.0001。CWTGRNN模型对20 个预测样本集中的樱桃含糖量的预测相对误差在2% 以内。结果表明,可见- 近红外光谱技术可以快速、准确和无损地测定樱桃中的含糖量,本研究提出的方法可以用于果蔬产业的品质管理与控制。     

近红外漫反射光谱技术对小米产地溯源的研究

为建立小米产地溯源的快速检测技术,更好的维护地方名优小米品牌效益,试验利用近红外漫反射光谱技术对不同状态小米进行产地溯源鉴别,试验分别选取来自肇源和肇州两个小米主产区的144份小米样品,应用近红外漫反射光谱技术结合化学计量学对不同状态下的小米进行产地溯源研究,结果表明:在全波长范围内采用因子化法建立的定性分析模型和在特征波段范围内采用偏最小二乘法(PLS)建立的定量分析模型,对肇源、肇州两个小米主产区的小米籽粒和小米粉末的正确鉴别率均在90%以上,其中小米粉末的模型正确预测率要高于小米籽粒。因此,应用近红外漫反射光谱技术对不同状态小米产地溯源的鉴别具有一定的可行性。     

近红外漫反射光谱检测损伤猕猴桃内部品质

研究贮藏期间损伤猕猴桃内部品质与其近红外漫反射光谱之间的关系。利用近红外光谱(12000～4000cm-1)技术和多元线性回归(multiple linear regression,MLR)、主成分回归(principal component regression,PCR)和偏最小二乘法(partial least squares,PLS)3种校正方法分别对损伤华优猕猴桃在2℃条件下贮藏4周期间的可溶性固形物含量、pH值和硬度进行定量分析;并对比吸光度原始光谱、一阶微分和二阶微分3种不同预处理方法的PLS模型校正结果。结果表明：一阶微分预处理方法时,应用PLS建立的可溶性固形物含量、pH值和硬度校正模型的效果最佳;预测集样品预测值与测量值之间的相关系数分别为0.812、0.703、0.919,预测均方根误差分别为0.749、0.153、1.700。说明应用近红外漫反射技术检测贮藏期间损伤猕猴桃的内部品质是可行的。     

近红外光谱技术在制糖业中的应用

近红外光谱（Near infraed spectroscopy简称MRS）分析技术是一种高效、快速的现代分析技术，己在很多领域得到广泛应用。对近红外光谱分析技术的发展历程、技术原理、分析程序、技术特点及其在制糖工业研究和应用作了简要介绍，并对其应用前景进行了讨论。     

Identification of animal meat muscles by visible and near infrared reflectance spectroscopy

Visible (VIS) and near infrared reflectance spectroscopy (NIRS) was used to identify and authenticate different meat muscle species. Samples from beef (n: 100), lamb (n: 140), pork (n: 44) and chicken (n: 48) muscles were homogenised and scanned in the visible (VIS) and near infrared (NIR) region (400-2500 nm) in a monochromator instrument in reflectance. Both Principal Component Analysis (PCA) and dummy partial least-squares regression (PLS) models were developed to identify different meat species. The models correctly classified more than 80% of the meat sample muscles according with the muscle specie. The results showed the potential of VIS and NIR spectra as an objective and rapid method for authentication and identification of meat muscle species.     

Identification of some wheat varieties by near infrared reflectance spectroscopy

Variety identification by electrophoresis is not applicable to routine control in industry. In the present work, the feasibility of near infrared reflectance (n.i.r.) analysis was investigated. Two hundred and two wheat samples including 66 samples of six known varieties were collected and their n.i.r. spectra were recorded. Spectral data were mathematically corrected in order to reduce the effect of granularity on n.i.r. spectra, then Principal Component Analysis and Multiple Discriminant Analysis (MDA) were applied to the corrected data. MDA allowed an efficient identification of the genetic origin of unknown samples: on a prediction set, 87% of samples were correctly identified. The computerised identification procedure needed less than 20 records for one sample. Further studies are necessary before recommending n.i.r. as a routine screening method.     

用遗传算法提取南疆红枣总糖的近红外光谱特征波长

本研究尝试利用近红外光谱技术测量红枣的总糖含量,针对采用偏最小二乘（PLS）法建立近红外光谱预测模型时波长筛选问题,提出用联合区间偏最小二乘法（si PLS）与遗传算法（GA）相结合的方法遗传联合区间偏最小二乘法（GA-si PLS）来提取近红外光谱特征区域和特征波长,提高模型预测精度的方法。结果表明:将全谱等分成20个子区间,用联合区间偏最小二乘法优选出4个特征子区间,在这4个子区间的基础上再用遗传偏最小二乘法继续筛选出12个特征波长。用12个特征波长建立的偏最小二成模型精度要好于全谱建立的模型,其主因子数减少了4个,预测集标准偏差（RMSECP）减少了25%,预测相关系数（RP）提高了5%。该方法选取的波长变量建立的校正模型,不仅使模型简洁、优化,而且增强了模型的预测能力。      

Application of reflectance near infrared spectroscopy for bread analyses

The feasibility of using near infrared spectroscopy (NIRS) for prediction of nutrients in a wide range of bread varieties mainly produced from wheat and rye was investigated. Calibration was performed on samples collected over a 3-year-period and the calibration equations were tested on samples collected the subsequent year. Bread samples were dried, crushed, ground and measured in a rotating sample cup in the wavelength range 1100–2500 nm. Full range or segmented reflectance data was used for calibration based on partial least square (PLS) regression. Protein, fat, dietary fibre, sugar, ash, saturated fatty acids (SFA), monounsaturated fatty acids (MUFA), polyunsaturated fatty acids (PUFA) and Na could be determined directly with r² values of 0.99, 0.99, 0.89, 0.96, 0.91, 0.90, 0.91, 0.92 and 0.76, respectively. The total contents of carbohydrates and energy was calculated from NIR data with r² values of 0.98 and 0.99. The ratios between analyte variation range standard deviation (SD) and the root mean square error of cross validation (RMSECV) were 8.3 (protein), 9.1 (fat), 3.0 (dietary fibre), 4.7 (sugar), 3.0 (ash), 3.1 (SFA), 3.3 (MUFA), 3.5 (PUFA), 1.9 (Na), 7.2 (carbohydrates) and 8.4 (energy). Equivalent ratios were obtained on an independent test set. It is concluded that the applied NIRS methodology is suitable for routine analysis of wheat and rye based bread for the investigated organic properties and ash. The technique may also give a rough estimate of the Na content.     

Discrimination of European wheat varieties using near infrared reflectance spectroscopy

Near infrared reflectance (NIR) spectroscopy combined with chemometrics was used to discriminate wheat varieties. A total of 249 samples of different wheat varieties from the 2003–2004 harvest were used to develop the best discriminant equation, by applying various scatters and mathematical treatments in the range of 400–2500 nm. Wheat varieties from Spain were ‘Sarina’, ‘Bolero’, ‘Berdún’, ‘Soisson’, ‘Chamorro’, ‘Artur Nick’, ‘Berdun’, ‘Marius’, ‘Anza’, ‘Kalifa’, and wheat varieties from France were ‘Galibier’ and ‘Quality’. The equation developed with the highest accuracy had an applied scatter of weighted multiplicative scatter correction, a math treatment of 2, 15, 8 (order of derivative, gap data points over which the derivative was taken, number of data points used in performing average smoothing). The percentage of correctly identified varieties was 99.5% for the calibration sample set and 94% for the validation sample set. The results demonstrated the usefulness of NIRS combined with chemometrics as a rapid method for discrimination of European wheat varieties. Although the application of the discriminant equation developed for the 2003–2004 harvest yielded a high rate, further test measurements are necessary to evaluate the robustness of the equation.     

近红外旋光法与普通旋光法测定蔗糖分的对比研究

就糖厂中间制品的蔗糖分用近红外旋光法和普通旋光法进行了测定,并且引入离子色谱法测定作为评判,对比分析,初步得到近红外旋光法较之普通旋光法的测定准确度要高的结论。     

Application of linear/non-linear classification algorithms in discrimination of pork storage time using Fourier transform near infrared (FT-NIR) spectroscopy

To address the rapid and nondestructive determination of pork storage time associated with its freshness, Fourier transform near infrared (FT-NIR) spectroscopy technique, with the help of classification algorithm, was attempted in this work. To investigate the effects of different linear and non-linear classification algorithms on the discrimination results, linear discriminant analysis (LDA), K-nearest neighbors (KNN), and back propagation artificial neural network (BP-ANN) were used to develop the discrimination models, respectively. The number of principal components (PCs) and other parameters were optimized by cross-validation in developing discrimination models. Experimental results showed that the performance of BP-ANN model was superior to others, and the optimal BP-ANN model was achieved when 5 PCs were included. The discrimination rates of the BP-ANN model were 99.26% and 96.21% in the training and prediction sets, respectively. The overall results sufficiently demonstrate that the FT-NIR spectroscopy technique combined with BP-ANN classification algorithm has the potential to determine pork storage time associated with its freshness.     

Real-time modeling of milk coagulation using in-line near infrared spectroscopy

This paper considers the extraction of meaningful information in real-time from near infrared (NIR) reflection measurements of coagulating milk. This information can be used for developing automatic cutting time determination. NIR spectra (1000-2500 nm) recorded during coagulation were compressed by principal component analysis. Using component scores as a function of time, two models are proposed for describing the three milk coagulation processes: κ-casein proteolysis, micelle aggregation, and network formation. A model for the entire coagulation process and a composite model for the three individual coagulation processes were established and tested on 12 cheese batches. Both models fitted very well (R² > 0.99) to the experimental NIR data. An algorithmic procedure is presented that is able to provide real-time parameter estimation for a semi-empirical model describing the kinetics of the milk coagulation processes as well as determining the transition times between the three different coagulation processes.     

Measurement of fat and moisture in air-dried bread by near infrared reflectance

Near infrared calibrations have been derived and used routinely for a year in the measurement of fat and moisture in air-dried bread. First and second derivative calibrations were obtained using a Pacific Scientific mark II scanning spectrophotometer on samples sent from all over South Africa to the Wheat Board for analysis. Prediction analysis performed on further bread samples gave standard errors of prediction (s.e.p.) of 0.12% fat and 0.13% moisture.