洋香瓜糖度檢測之研究－（二）近紅外線分析法

In a glycolipid fermentation, mannosyl erythritol lipid (MEL) is produced from soybean oil added to a medium as a source of carbon. A measurement system for the concentrations of MEL and soybean oil in the fermentation process has been developed using near infrared (NIR) spectroscopy. MEL and soybean oil in the culture broth were extracted with ethyl acetate. NIR spectra of the ethyl acetate extract were measured in the wavelength range between 400 and 2500 nm at 2 nm intervals. The absorption caused by MEL was observed at 1436, 1920 and 2052 nm. To obtain a calibration equation, a multiple linear regression (MLR) was carried out between the second derivative NIR spectral data at 2040 and 1312 nm and MEL concentrations obtained using thin-layer chromatography with a flame-ionisation detector (TLC/FID) method. The values of the regression coefficient ( R) and the standard error of calibration ( SEC) were 0.994 and 0.48 g L−1, respectively. The absorption caused by soybean oil was observed at 1208, 1716, 1766, 2182 and 2302 nm. A calibration equation for soybean oil was formulated with the second derivative NIR spectral data at 2178 and 2090 nm. The values of R and SEC were 0.974 and 0.77 g L−1, respectively. After validation of the calibration equation, good agreement was observed between the results of the TLC/FID method and those of the NIR method for both constituents. The values of the correlation coefficient ( r) for MEL and the standard error of prediction ( SEP) were 0.994 and 0.45 g L−1, respectively. The values of r and SEP for soybean oil were 0.979 and 0.56 g L−1, respectively. The NIR method was applied to the measurement of the concentrations of MEL and soybean oil in an actual fermentation process and good results were obtained. The study indicates that NIR spectroscopy is a useful method for the measurement of the raw material and product in glycolipid fermentation.

To develop a nondestructive sugar analyzer for intact peaches, the potential of the near infrared (NIR) method with fiber optics in interactance mode was evaluated. In 'Shimizu Hakuto', a white peach cultivar, there were clear differences among the NIR second derivative spectra of high, medium, and low Brix values in the wavelength region around 906 nm. This wavelength was assigned to sucrose and selected as first calibration wavelength. The highest multiple correlation coefficient of 0.97 was obtained in the linear regression equation that included four optical terms of d2log(1/R) at 906, 878, 870 and 889 nm, while a standard error of calibration (SEC) was 0.48°Brix. A bias-corrected standard error of prediction (SEP) was 0.50°Brix with a bias of 0.01°Brix, suggesting that the MR method is sufficiently accurate to determine sugar content in intact peaches to be practical.

Grass (monocots) and non-grass (dicots) proportions in ruminant diets are important nutritionally because the non-grasses are usually higher in nutritive value, particularly protein, than the grasses, especially in tropical pastures. For ruminants grazing tropical pastures where the grasses are C4 species and most non-grasses are C3 species, the ratio of 13C/12C in diet and faeces, measured as δ13C‰, is proportional to dietary non-grass%. This paper describes the development of a faecal near infrared (NIR) spectroscopy calibration equation for predicting faecal δ13C from which dietary grass and non-grass proportions can be calculated. Calibration development used cattle faeces derived from diets containing only C3 non-grass and C4 grass components, and a series of expansion and validation steps was employed to develop robustness and predictive reliability. The final calibration equation contained 1637 samples and faecal δ13C range (‰) of [12.27]–[27.65]. Calibration statistics were: standard error of calibration (SEC) of 0.78, standard error of cross-validation (SECV) of 0.80, standard deviation ( SD) of reference values of 3.11 and R2 of 0.94. Validation statistics for the final calibration equation applied to 60 samples were: standard error of prediction ( SEP) of 0.87, bias of −0.15, R2 of 0.92 and RPD of 3.16. The calibration equation was also tested on faeces from diets containing C4 non-grass species or temperate C3 grass species. Faecal δ13C predictions indicated that the spectral basis of the calibration was not related to 13C/12C ratios per se but to consistent differences between grasses and non-grasses in chemical composition and that the differences were modified by photosynthetic pathway. Thus, although the calibration equation could not be used to make valid faecal δ13C predictions when the diet contained either C3 grass or C4 non-grass, it could be used to make useful estimates of dietary non-grass proportions. It could also be used to make useful estimates of non-grass in mixed C3 grass/non-grass diets by applying a modified formula to calculate non-grass from predicted faecal δ13C. The development of a robust faecal-NIR calibration equation for estimating non-grass proportions in the diets of grazing cattle demonstrated a novel and useful application of NIR spectroscopy in agriculture.

Comparison of several variable selection methods for quantitative analysis and monitoring of the Yangxinshi tablet process using near-infrared spectroscopy

Monitoring of methanogen density using near-infrared spectroscopy

Five Populus × euramericana wood samples representing three different sites were selected and near infrared (NIR) spectra were obtained. For these sections, basis weight, brightness and three mechanical properties (tensile index, tearing index and bursting index) were determined by standard analytical methods. Calibrations were developed for each paper property using the NIR spectra, data on paper properties, using partial least squares (PLS) regression. The results show that the coefficients of correlation of calibration and validation for basis weight were 0.8824 and 0.8299, respectively; the standard error of calibration (SEC) and prediction (SEP) were 1.150 and 1.170, respectively. In testing for brightness, the correlation coefficient of calibration was 0.9621 and for validation 0.9612, while the SEC and SEP were 0.997 and 1.300, respectively; paper brightness and NIR spectroscopy were highly correlated. NIR spectroscopy can be used to predict tensile, tearing and bursting indices of paper samples rapidly. We found that the paper properties fitted by NIR and GB methods were highly correlated. The coefficients of correlation of calibration and validation for basis weight exceeded 0.8000, while the SEC and SEP were very small. These results reveal that the five paper properties of Populus × euramericana and those predicted by the NIR model were highly correlated. We conclude that the NIR models can be used for the prediction of paper properties.

The sodium chloride content in the flesh of tuna fish is one of the factors for determining the price in the fishing industry. Titration is a standard method for the analysis of salt and this is time consuming. Near infrared spectroscopy is a potential alternative method for rapid detection without the need for wet chemical assay. Although sodium chloride is infrared inactive, this study investigated the influence of salt on the absorbance of near infrared energy and showed that the sodium chloride content can be determined using changes in the water band at 970 nm. Calibration equations were developed from frozen fish pieces and ground samples using multiple linear regression for the wavelength region of 700–1000 nm. The best result was achieved from frozen samples with a coefficient of determination for the calibration set ([Formula: see text]) = 0.71, standard error of calibration (SEC) = 0.20%, coefficient of determination for the validation set ([Formula: see text]) = 0.64, standard error of prediction (SEP) = 0.26% and bias = − 0.00%. In order to verify the significant variables used to determine infrared inactive sodium chloride, partial least squares regression was performed on frozen samples. The important variable in multiple linear regression and partial least squares regression was the absorbance band at 976 nm attributed to water molecules. The result from partial least squares calibration showed [Formula: see text] = 0.83, SEC = 0.20%, [Formula: see text] = 0.54, SEP = 0.25% and bias = 0.00%. The salt values predicted using the near infrared models were not significantly different from the reference values obtained by the standard titration method at the 95% confidence interval.

Quantification of glycine crystallinity by near-infrared (NIR) spectroscopy

This work aimed to prove the feasibility of using near infrared (NIR) spectroscopy to detect vegetable proteins (soy, pea and wheat isolates) in milk powder. One hundred and fifty-five samples of genuine and adulterated milk powder (NIZO, Ede, The Netherlands) were analysed by NIR spectroscopy using an InfraAlyzer 500 (Bran+Luebbe). NIR spectra were collected at room temperature, and data were processed by Sesame Software (Bran+Luebbe) to select the most significant wavelengths. Calibrations were made on all samples in the range 0–5% of added vegetable proteins. Separate prediction and validation sets were prepared using samples not included in the calibration set. The best results were obtained by applying multiple linear regression (MLR) to the first derivative of absorbance values at five wavelengths. NIR spectroscopy proved a very effective tool for detecting the presence of vegetable proteins in milk powder on the basis of standard error of calibration ( SEC), standard error of prediction ( SEP) and standard error of validation ( SEV) values found in this study. The calibration carve was characterised by a regression coefficient ( R2) equal to 0.993 with a SEC of 0.20, a SEP of 0.23 and a SEV of 0.21.

ABSTRACTA novel method was developed for the quality control of Ephedrae herba by near-infrared (NIR) spectroscopy. First, qualitative models established by discriminant analysis and support vector machine were used for the preliminary screening of unqualified samples of E. herba. Then quantitative models of ephedrine and the total alkali (ephedrine and pseudoephedrine) were established by partial least squares regression and particle swarm optimization based least square support vector machine. The contents of test samples were predicted by the established NIR quantitative models. As a result, the accuracies of unqualified identification were 98.9% by discriminant analysis and 100% by support vector machine. The performance of the particle swarm optimization based least square support vector machine models were better than the partial least squares regression models. The correlation coefficients were both more than 0.98 and relative standard errors of calibrations were less than 9% in the calibration sets of particle swarm optimization based least square support vector machine models. As for the test sets, the correlation coefficients were both more than 0.93 and the relative standard errors of prediction were less than 13%, indicating satisfactory predicted results. All of these results demonstrated that NIR spectroscopy may be a powerful tool for the quality control of E. herba.

Near-infrared reflectance spectroscopy (NIRS) was used to estimate N, neutral detergent fibre (NDF), acid detergent fibre (ADF), lignin and cellulose contents in leaves of a heterogeneous group of 17 woody species from the Central Western region of the Iberian Peninsula. The sample set consisted of 182 samples of leaves of deciduous and evergreen species, showing a wide range of concentrations determined by reference methods: 6.60-35.2 g kg-1 (N), 15.5-66.0% (NDF), 10.2-57.3% (ADF), 3.45-27.4% (lignin) and 5.79-31.3% (cellulose). Reflectance spectra, obtained for samples of dried and ground leaves, were recorded as log1/R (R=reflectance) from 1,100 to 2,500 nm. NIRS calibrations were developed using multiple linear (MLR) and partial least-squares (PLSR) regressions, and tested by external validation. Spectral data were transformed to the first and second derivative (1D, 2D). The PLSR method and derivative transformations provided the best statistics and showed lower standard errors of calibration (SEC) and higher coefficients of multiple determination (R2). In the external validation the standard errors of prediction (SEP) were 0.76 g kg-1 (N), 2.11% (NDF), 1.47% (ADF), 0.85% (lignin) and 0.86% (cellulose). The results obtained show that NIRS is very effective for the estimation of these organic constituents in leaf tissue of woody species. This technique can be used in ecological or ecophysiological studies as an alternative to the more time-consuming standard methods.

Near infrared analysis was used to predict the starch, moisture, and sugar content in sliced fresh sweetpotato [ Ipomoea batatas (L.) Lam.] storage roots. Samples were collected in each of three growing years. The best calibration equation for starch from combined samples (1989 to 1991) showed a multiple correlation coefficient ( R ) of 0.949, a standard error of calibration ( sec ) of 2.01, and a standard error of prediction ( sep ) of 1.91. The R , sec , and sep for moisture and sugar were 0.930, 1.85, and 2.00, and 0.837, 1.30, and 1.21, respectively. Calibrations based on samples from a given year adequately predicted the variables but could not account for variances introduced by samples from other years. Multiyear calibrations based on several years of data adequately predicted starch and moisture content in root slices. Thus, multiyear calibrations with annual bias adjustments can be applied to screening sweetpotato breeding germplasm for these two variables.

Quantitative real-time release testing of rhubarb based on near-infrared spectroscopy and method validation

Eight Pinus radiata D. Don (Radiata pine) increment core samples representative of a total of thirty-two increment cores were selected. NIR spectra were obtained from the radial–longitudinal face of each core in 10-mm increments and used to develop a microfibril angle (MFA) calibration. The MFA calibration was developed using seven factors giving an excellent relationship between SilviScan-2 determined MFA and NIR fitted MFA (coefficient of determination (R2) = 0.95) and a standard error of calibration (SEC) of 1.8 degrees. The MFA calibration was used to predict the MFA of NIR spectra obtained in 10-mm sections from the radial–longitudinal face of two intact P. radiata increment cores. NIR predicted MFA was found to be in excellent agreement with MFA determined by SilviScan-2, with R2 of 0.98 (core A) and 0.96 (core B). The standard error of prediction (SEP) for core A (1.0 degree) was much lower than for core B (2.5 degrees). Both sets of predictions closely followed the patterns of MFA radial variation determined by SilviScan-2. NIR spectroscopy provides a rapid method for determining MFA variation in increment cores and is suitable for the routine analysis of large numbers of samples.

Quantitation of protein, creatinine, and urea in urine by near-infrared spectroscopy