Prediction of enological parameters and discrimination of rice wine age using least-squares support vector machines and near infrared spectroscopy

The use of least-squares support vector machines (LS-SVM) combined with near-infrared (NIR) spectra for prediction of enological parameters and discrimination of rice wine age is proposed. The scores of the first ten principal components (PCs) derived from PC analysis (PCA) and radial basis function (RBF) were used as input feature subset and kernel function of LS-SVM models, respectively. The optimal parameters, the relative weight of the regression error gamma and the kernel parameter sigma 2, were found from grid search and leave-one-out cross-validation. As compared to partial least-squares (PLS) regression, the performance of LS-SVM was slightly better, with higher determination coefficients for validation ( Rval2) and lower root-mean-square error of validation (RMSEP) for alcohol content, titratable acidity, and pH, respectively. When used to discriminate rice wine age, LS-SVM gave better results than discriminant analysis (DA). On the basis of the results, it was concluded that LS-SVM together with NIR spectroscopy was a reliable and accurate method for rice wine quality estimation.     

滩涂土壤有机质含量的反射光谱估算

Rapid determination of soil organic matter (SOM) using regression models based on soil reflectance spectral data serves an important function in precision agriculture. “deviation of arch”(DOA)-based regression and partial least squares regression (PLSR) are two popular modeling approaches to predict SOM. However, few studies have explored the accuracy of the DOA-based regression and PLSR models. Therefore, the DOA-based regression and PLSR were applied to the visible near-infrared (VNIR) spectra to estimate SOM content in the case of various dataset divisions. A two-fold cross-validation scheme was adopted and repeated 10 000 times for rigorous evaluation of the DOA-based models in comparison with the widely used PLSR model. Soil samples were collected for SOM analysis in the coastal area of northern Jiangsu Province, China. The results indicated that both modelling methods provided reasonable estimates of SOM, with PLSR outperforming DOA-based regression in general. However, the performance of PLSR for the validation dataset decreased more noticeably. Among the four DOA-based models, the linear model of the DOA provided the best estimation of SOM and a cutoff of SOM content (19.76 g kg^-1), and the performance for calibration and validation datasets was consistent. As the SOM content exceeded 19.76 g kg^-1, SOM became more effective in masking the spectral features of other soil properties to a certain extent. This work confirmed that reflectance spectroscopy combined with PLSR could serve as a non-destructive and cost-efficient way for rapid determination of SOM when hyperspectral data were available. The DOA-based model, which requires only 3 bands in the visible spectra, also provided SOM estimation with acceptable accuracy.     

NMR characterization of saccharides in Italian honeys of different floral sources

The saccharide profiles of 5 different botanical species in 86 Italian honey samples were investigated by 1H and 1H-13C NMR spectroscopy. Nineteen saccharides were identified in the aqueous extracts, namely, fructose, glucose, gentiobiose, isomaltose, kojibiose, maltose, maltulose, melibiose, nigerose, palatinose, sucrose, turanose, erlose, isomaltotriose, kestose, maltotriose, melezitose, raffinose, and maltotetraose. PCA performed on NMR spectral regions, in particular between 4.400 and 5.700 ppm and the fructose signal at 4.050 ppm, revealed a partial sample grouping. The score contribution plots derived from PCA performed using the mean values for the buckets of the anomeric region for each floral source allowed the identification of saccharides characterizing different honeys. OPLS-DA models were further evaluated to confirm the previous findings. OPLS-DA models were also built to highlight differences between polyfloral and high mountain polyfloral honeys and between high mountain polyfloral and rhododendron honeys, both collected at high altitude; S-plots highlighted the characteristic saccharides.     

A Comparative Study of Fourier Transform Raman and NIR Spectroscopic Methods for Assessment of Protein and Apparent Amylose in Rice

Fourier‐transform Raman (FT‐Raman) spectroscopy and near‐infrared (NIR) reflectance spectroscopy were used to compare calibration models for determining rice cooking quality parameters such as apparent amylose and protein. Samples from two seasons were used in each calibration set. The laboratory values ranged from 4.89 to 12.48% for protein and from 0.2 to 25.7% for amylose. The data for both FT‐Raman and NIR were preprocessed with orthogonal signal correction (OSC) for standardization. For both spectroscopic methods, five models were optimized by partial least squares regression (PLSR) and by Martens' uncertainty regression (MUR), including no processing, smoothing, normalization, first derivative (D1), and second derivative (D2). Based solely on standard error of cross‐validation (SECV), the FT‐Raman method was superior to the NIR method for protein. For amylose, the FT‐Raman and NIR methods resulted in similar calibration statistics with a high precision, with the FT‐Raman requiring fewer factors. The best FT‐Raman models were generated from OSC preprocessing with MUR for protein (SECV 0.15%, five factors) and from OSC without MUR for amylose (SECV 0.70%, seven factors). The best NIR models were obtained with D2 transform of OSC spectra for protein (SECV 0.22%, four factors) and with OSC spectra for amylose (SECV 0.57%, 11 factors).     

Geographic classification of spanish and Australian tempranillo red wines by visible and near-infrared spectroscopy combined with multivariate analysis

Visible (vis) and near-infrared (NIR) spectroscopy combined with multivariate analysis was used to classify the geographical origin of commercial Tempranillo wines from Australia and Spain. Wines (n = 63) were scanned in the vis and NIR regions (400-2500 nm) in a monochromator instrument in transmission. Principal component analysis (PCA), discriminant partial least-squares discriminant analysis (PLS-DA) and linear discriminant analysis (LDA) based on PCA scores were used to classify Tempranillo wines according to their geographical origin. Full cross-validation (leave-one-out) was used as validation method when PCA and LDA classification models were developed. PLS-DA models correctly classified 100% and 84.7% of the Australian and Spanish Tempranillo wine samples, respectively. LDA calibration models correctly classified 72% of the Australian wines and 85% of the Spanish wines. These results demonstrate the potential use of vis and NIR spectroscopy, combined with chemometrics as a rapid method to classify Tempranillo wines accordingly to their geographical origin.     

Gas-liquid chromatographic test for honey adulteration by high fructose corn sirup.

A gas-liquid chromatographic (GLC) method has been developed to detect the addition to honey of high fructose corn sirup (HFCS). Samples are derivatized directly with trimethylsilane, cholestane is added as an internal standard, and the levels of maltose (includes other minor disaccharides) and isomaltose are determined after chromatography on OV-17. Domestic and imported honey samples (115) contained 2.00% maltose and 0.71% isomaltose. HFCS samples (21) contained 1.50% maltose, and 2.09% isomaltose. A discriminatory equation was developed (D = 2.73 - 5. 35 (Isomaltose/maltose)) and, when applied to the data for these samples and 37 adulterated samples, 81.4% of authentic honey samples and 78.4% of samples known to be adulterated with HFCS were correctly classified.     

Development of a Near‐Infrared Imaging System for Determination of Rice Moisture

The objective of this study was to develop a near‐infrared (NIR) imaging system to determine rice moisture content. The NIR imaging system fitted with 15 band‐pass filters (wavelengths of 870–1,014 nm) was used to capture the spectral image. In this work, calibration methods including multiple linear regression (MLR), partial least squares regression (PLSR), and artificial neural network (ANN) were used in both near‐infrared spectrometry (NIRS) and the NIR imaging system to determine the moisture content of rice. Comprehensive performance comparison among MLR, PLSR, and ANN approaches has been conducted. To reduce repetition and redundancy in the input data and obtain a more accurate network, six significant wavelengths selected by the MLR model, which had high correlation with the moisture content of rice, were used as the input data of the ANN. The performance of the developed system was evaluated through experimental tests for rice moisture content. This study adopted the coefficient of determination (r_val²), the standard error of prediction (SEP), and the relative performance determinant (RPD) as the performance indices of the NIR imaging system with respect to the tests of rice moisture content. Utilizing these three models, the analysis results of r_val², SEP, and RPD for the validation set were within 0.942–0.952, 0.435–0.479%, and 4.2–4.6, respectively. From experimental results, the performance of NIR imaging system was almost the same as that of NIRS. Using the developed NIR imaging system, all of the three different calibration methods (MLR, PLSR, and ANN) provided a high prediction capacity for the determination of moisture in rice samples. These results indicated that the NIR imaging system developed in this study can be used as a device for the measurement of rice moisture content.     

基于氨基酸组成的黄酒酒龄电子舌鉴别

该研究采用电子舌结合化学计量学方法用于黄酒酒龄的快速鉴别。为确证黄酒样品酒龄,采用氨基酸分析仪分析了1年陈、3年陈和5年陈黄酒中20种氨基酸,并利用主成分分析（principal component analysis,PCA）对氨基酸数据进行了分析。采用电位型电子舌采集了不同酒龄黄酒样品的味觉指纹信息,并采用判别分析（discriminant analysis,DA）方法结合味觉指纹信息建立黄酒酒龄快速鉴别模型。采用偏最小二乘法（partial least squares regression,PLSR）建立电子舌响应信号与氨基酸含量之间的相关关系。氨基酸数据结合PCA分析表明所有样品均标注正确;电子舌结合DA所建黄酒酒龄鉴别模型可将3个年份预测集样品正确区分;异亮氨酸（Ile）、天门冬氨酸（Asp）、酪氨酸（Tyr）和缬氨酸（Val）与电子舌相关性高,模型的相对分析误差（Residual predictive deviation, RPD）高于2。研究表明电位型电子舌结合判别分析是黄酒龄鉴别的稳健方法。     

Measurement of condensed tannins and dry matter in red grape homogenates using near infrared spectroscopy and partial least squares

Samples (n = 620) of homogenized red grape berries were analyzed using a visible and near-infrared (NIR) spectrophotometer (400-2500 nm) in reflectance. The spectra and the analytical data were used to develop partial least-squares calibrations to predict dry matter (DM) content and condensed tannins (CT) concentrations. The coefficient of determination in cross-validation and the standard error of cross-validation were 0.92 and 0.83% w/w for DM and 0.86 and 0.46 mg/g epicatechin equivalents for CT, respectively. The standard error in prediction was 1.34% w/w for DM and 0.89 mg/g epicatechin equivalents for CT, respectively. By implementing a NIR spectroscopy method to measure DM and CT in red grape homogenates, we have developed an approach that is suited to large-scale compositional analysis in commercial wine production facilities, as it enables the analysis of large numbers of samples needed to stream batches of fruit. From an economical point of view, the calibration models could be achieved with relatively small data sets. Thus, NIR offers a suitable and efficient tool for the simultaneous measurement of DM and CT in addition to other important parameters in red grape homogenates such as total anthocyanins, total soluble solids, and pH, with minimal sample preparation and low cost.     

近红外透反射光谱测定单粒稻种的蛋白质含量

为探索NIR光谱技术在水稻种子蛋白质含量分析中的应用,本研究细致分析了单粒稻种在不同光谱采集方式下的近红外光谱(NIRS)特征,并利用离子束诱变育种得到的水稻9311突变体库的种子,建立准确性较好的单粒糙米和单粒稻种的蛋白质定量模型。结果表明,与漫反射光谱采集方式下的单粒糙米蛋白质模型相比,透反射和透射光谱采集方式下能得到相关性较好的糙米蛋白质模型,其中单粒糙米蛋白质最优定量模型的决定系数(R²)为0.97,预测均方根误差(RMSEP)为0.27%。在单粒稻种中,由于种壳的反射作用,漫反射光谱采集方式下依然无法建立准确性高的蛋白质模型,透反射光谱采集方式下能够建立具有一定预测能力的蛋白质定量模型(RMSEP=0.81%),透射光谱采集方式下能够建立准确性高的蛋白质定量模型(R²=0.96,RMSEP=0.24%)。本研究结果为无损快速分析单粒稻种提供了一种解决方法。     

Rapid prediction of gross energy and utilizable energy in cereal food products using near-infrared reflectance spectroscopy

Near-infrared (NIR) spectroscopy has been used in foods for the rapid assessment of several macronutrients; however, little is known about its potential for the evaluation of the utilizable energy of foods. Using NIR reflectance spectra (1104-2494 nm) of ground cereal products (n = 127) and values for energy measured by bomb calorimetry, chemometric models were developed for the prediction of gross energy and available energy of diverse cereal food products. Standard errors of cross-validation for NIR prediction of gross energy (range = 4.05-5.49 kcal/g), energy of samples after adjustment for unutilized protein (range = 3.99-5.38 kcal/g), and energy of samples after adjustment for unutilized protein and insoluble dietary fiber (range = 2.42-5.35 kcal/g) were 0.053, 0.053, and 0.088 kcal/g, respectively, with multiple coefficients of determination of 0.96. Use of the models on independent validation samples (n = 58) gave energy values within the accuracy required for U.S. nutrition labeling legislation. NIR spectroscopy, thus, provides a rapid and accurate method for predicting the energy of diverse cereal foods.     

Near-Infrared Spectroscopy Coupled with Chemometrics Tools: A Rapid and Non-Destructive Alternative on Soil Evaluation

Near-infrared (NIR) spectroscopy is a rapid, non-destructive and accurate technique for analyzing a wide variety of samples, thus, the growing interest of using this technique in soil science. The objective of this study was to evaluate the potential of NIR spectroscopy to predict organic carbon (OC), total nitrogen (TN), available phosphorus (P) and available potassium (K) in the soil. NIR spectra from 20 cm³ of soil samples were acquired on the range of 750 to 2500 nm in diffuse reflectance mode, resolution of 16 cm^?1 and 64 scans. Eight models of calibration/validation were constructed. Calibration and validation models showed that the predictive potential of NIR varied with the specific soil property (OC, TN, P and K) under evaluation and according to the methodology employed in the model construction (cross-validation or test set). Good prediction models were obtained for OC and TN content based on the statistical parameters. Test set methodology was able to predict soil OC, TN, P, and K better than cross-validation methodology.     

Influence of the number of samples on prediction error of visible and near infrared spectroscopy of selected soil properties at the farm scale

This paper reports on the influence of the number of samples used for the development of farm‐scale calibration models for moisture content (MC), total nitrogen (TN) and organic carbon (OC) on the prediction error expressed as root mean square error of prediction (RMSEP) for visible and near infrared (vis‐NIR) spectroscopy. Fresh (wet) soil samples collected from four farms in the Czech Republic, Germany, Denmark and the UK were scanned with a fibre‐type vis‐NIR, AgroSpec spectrophotometer with a spectral range of 305–2200 nm. Spectra were divided into calibration (two thirds) and prediction (one third) sets and the calibration spectra were subjected to a partial least squares regression (PLSR) with leave‐one‐out cross‐validation using Unscrambler 7.8 software. The RMSEP values of models with a large sample number (46–84 samples from each farm) were compared with those of models developed with a small sample number (25 samples selected from the large sample set of each farm) for the same variation range. Both large‐set and small‐set models were validated by the same prediction set for each property. Further PLSR analysis was carried out on samples from the German farm, with different sample numbers of the calibration set of 25, 50, 75 and 100 samples. Results showed that the large‐size dataset models resulted in smaller RMSEP values than the small‐size dataset models for all the soil properties studied. The results also demonstrated that with the increase in sample number used in the calibration set, RMSEP decreased in almost linear fashion, although the largest decrease was between 25 and 50 samples. Therefore, it is recommended that the number of samples should be chosen according to the accuracy required, although 50 soil samples is considered appropriate in this study to establish calibration models of TN, OC and MC with smaller expected prediction errors as compared with smaller sample numbers.     

Evaluation of nonstarch polysaccharides and oligosaccharide content of different soybean varieties (Glycine max) by near-infrared spectroscopy and proteomics

A total of 832 samples of soybeans were screened by near-infrared (NIR) reflectance spectroscopy, to identify soybean samples with a lower content of oligosaccharides and nonstarch polysaccharides (NSP). Of these, 38 samples were identified on the basis of variation in protein content and agronomic value and submitted to high-resolution NIR spectroscopy. On the basis of the NIR data, 12 samples were further selected for chromatographic characterization of carbohydrate composition (mono-, di-, and oligosaccharides and NSP). Their soluble proteins were separated by two-dimensional gel electrophoresis (2DE). Using partial least-squares regression (PLSR), it was possible to predict the content of total NSP from the high-resolution NIR spectra, suggesting that NIR is a suitable and rapid nondestructive method to determine carbohydrate composition in soybeans. The 2DE analyses showed varying intensities of several proteins, including the glycinin G1 precursor. PLSR analysis showed a negative correlation between this protein and insoluble NSP and total uronic acid (UA).     

Classification of wine and alcohol vinegar samples based on near-infrared spectroscopy. Feasibility study on the detection of adulterated vinegar samples

Near-infrared (NIR) spectroscopy was used to discriminate between wine vinegar (red or white) and alcohol vinegar. One orthogonal signal correction method (OSC) was applied on a set of 73 vinegar NIR spectra from both origins and artificial blends made in the laboratory in order to remove information unrelated to a specific chemical response (tartaric acid), which was selected due to its high discriminant ability to differentiate between wine vinegar and alcohol vinegar samples. These corrected NIR spectra, as well as raw NIR spectra and 14 physicochemical variables, were used to develop separate classification models using the potential functions method as a class-modeling technique. The aforementioned models were compared to evaluate the suitability of NIR spectroscopy as a rapid method for discriminating between vinegar origins. The transformation of vinegar NIR spectra by means of an orthogonal signal correction method prompted a notable improvement in the specificity of the constructed classification models. The classification model developed was then applied to artificial vinegar blends made in the laboratory to test its capacity to recognize adulterated vinegar samples.     

Investigation of the potential utility of single-bounce attenuated total reflectance Fourier transform infrared spectroscopy in the analysis of distilled liquors and wines

A new Fourier transform infrared (FTIR) spectroscopic method based on single-bounce attenuated total reflectance (SB-ATR) spectroscopy was developed for the analysis of distilled liquors and wines. For distilled liquors, a partial least-squares (PLS) calibration was developed for alcohol determination based on the SB-ATR/FTIR spectra of mixtures of ethanol and distilled water. An independent set of 12 different distilled liquor samples was predicted from the PLS calibration, and a standard deviation of the differences for accuracy (SDD(a)) between actual and predicted values of 0.142% (v/v) was obtained. The potential utility of SB-ATR/FTIR spectroscopy for the analysis of wines was initially evaluated based on a comparison with Fourier transform near-infrared (FT-NIR) spectroscopy and FTIR spectroscopy using a flow-through transmission cell. PLS calibrations for alcohol, total reducing sugars, total acidity and pH were developed using pre-analyzed wine samples (n = 28), and for SB-ATR/FTIR spectroscopy, the SDD(a) for the leave-one-out cross-validation statistics were of the order of 0.100% (v/v), 0.707 g L(-1), 0.189 g L(-1) (H2SO4), and 0.230, respectively. Overall, the SB-ATR/FTIR results were better than those obtained using FT-NIR spectroscopy and comparable to those obtained with transmission FTIR spectroscopy. A PLS calibration based on preanalyzed wine samples (n = 72) for the prediction of 11 different components and parameters in wines by SB-ATR/FTIR spectroscopy was subsequently developed and validated using an independent sample set (n = 77). Good coefficients of correlation between the reference and predicted values for the validation set were obtained for most of the components and parameters except citric acid, volatile acids, and total SO2. The results of this study demonstrate the suitability of SB-ATR/FTIR spectroscopy for the routine analysis of distilled liquors and wines.     

Analysis of Fatty Acids in Kernel,Flour, and Oil Samples of Maize by NIR Spectroscopy Using Conventional Regression Methods

High cost and painstaking procedures associated with fatty acid analyses of maize kernel necessitate the use of alternative methods. NIR spectroscopy offers advantages in this respect for a variety of areas such as plant breeding, food and feed industries, and biofuel production, in which different forms of maize kernel (e.g., intact kernel, flour, or oil) are used as material. We investigated the possibility of estimating maize oil quality traits by using different samples (intact kernel, flour, and oil) and conventional regression methods (multiple linear regression [MLR] and partial least squares regression [PLSR]) applied to their NIR spectra. MLR and PLSR calibration models were developed for oleic acid, linoleic acid, oleic/linoleic acid ratios, total monounsaturated fatty acid, total polyunsaturated fatty acid (PUFA), and total saturated fatty acid by analyzing 120 maize samples. Robustness in terms of prediction accuracy of the models developed here was tested with a reserved set of samples (n = 30). The results suggested that fatty acids could be possibly estimated by calibrations developed from flour and oil samples with a high degree of accuracy, whereas intact samples did not offer satisfactory results. PLSR and MLR methods gave better results in flour and oil samples, respectively. PUFA was the trait that was most successfully estimated from both flour (for the PLSR model, standard error of the estimate [SEP] of 1.78%, relative performance to deviation [RPD] of 3.09, R² = 0.93) and oil (for the MLR model, SEP of 0.85%, RPD of 6.52, R² = 0.98) samples. We concluded that sample type and chemometric method should be handled as important factors in calibration development, and the effects of these factors may vary depending on the trait being analyzed.     

Quantification of saccharides in multiple floral honeys using fourier transform infrared microattenuated total reflectance spectroscopy

Fourier transform infrared (FTIR) spectroscopy with microattenuated total reflectance (mATR) sampling accessory and chemometrics (partial least squares and principal component regression) was used for the simultaneous determination of saccharides such as fructose, glucose, sucrose, and maltose in honey. Two calibration models were developed. The first model used a set of 42 standard mixtures of fructose, glucose, sucrose, and maltose prepared over the range of concentrations normally present in honey, whereas the second model used a set of 45 honey samples from various floral and regional sources. The developed models were validated with different data sets and verified by high-performance liquid chromatography (HPLC) measurements. The R (2) values between the FTIR-mATR predicted and HPLC results of the different sugars were between 0.971 and 0.993, demonstrating the predictive ability and accuracy of the procedure.     

可见/近红外光谱技术无损检测果实坚实度的研究

该研究的目的是建立可见/近红外光谱与梨果实坚实度之间的数学模型,评价可见/近红外光谱技术无损测量梨果实坚实度的应用价值.在可见/近红外光谱区域(350～1800nm),试验对比分析了不同测量部位、不同光谱预处理方法和不同校正建模算法的梨果实坚实度校正模型.结果表明:赤道部位吸光度一阶微分光谱的偏最小二乘回归所建梨果实坚实度校正模型的预测性能较优,其校正和预测相关系数分别为0.8779和0.8087,校正和预测均方误差分别为1.0804N和1.4455N.研究表明:可见/近红外光谱技术无损检测梨果实坚实度是可行的.