| Chinese rice wine is the most ancient unique wine kind of nutrition and health in China, and be honored as "essence" of the country. However, at present, the basic research towards rice wine is relative weak. The detailed changes in various flavor components during rice wine aging still remain uncertain. In additional, due to the backward technologies used in quality control of rice wine, only a few routine chemical parameters, such as alcohol degree, total sugars, total acids, were analyzed in conventional testes. Little is known about the small amounts of functional components that are closely related to its flavor, which cannot meet consumers’higher requirements towards rice wine products. It also remains as a barrier for the growth of rice wine industry. This study is a further extension of the former research projects carried out by our group members (flavor components quantification and age discrimination of rice wines based on NIR spectroscopy), with its aim to establish analytical methods and devices based on fingerprint techniques for aging status characterization as well as quality control and risk assessment of rice wine products.The object of this research is Shaoxing rice wine, which is the best representation of Chinese rice wine. Multiple technologies based on instrument analysis, VIS/NIR, NIR and MIR spectroscopy were applied as fingerprint techniques for quality evaluation of rice wine. Fingerprints of rice wine of various ageing status based on chemical compositions as well as NIR spectra were obtained. Quantification of flavor components in rice wine based on both NIR and MIR spectroscopy was also undertaken and compared. The effect of temperature on the performance for rice wine quality determination based on VIS/NIR spectroscopy was investigated, and the optimal temperature level was determined. A hardware and software system for quality determination of rice wine products based on temperature control was then developed.The main contents and conclusions were:(1) Continuous tracing of the changes in various flavor components in Shaoxing wines during the aging period of3to45months was undertaken. The results showed that,①Among six routine parameters, the content of alcohol degree and total sugar decreased, while the content of non-sugar solid and total acid increased. The change of pH and amino acid nitrogen was not clear. Besides pH, the remaining five indexes were all found to be significant different.②The content of lactic acid increased, similar to total acid. Significant difference was also detected.③Among five sugars, the change of glucose was not obvious, and no significant difference between different aging status was detected. The remaining four sugar parameters were all significant different, but their changes were also not obvious.④The content of Na, Mg, Al, K, Ca, Mn, Fe, Cu and Zn in rice wines fluctuated during aging, no obvious increase or decrease trend was observed. Besides Ca and Cu, other elements were all found to be significantly modified during ageing.⑤Among16free amino acids, the content of aspartic acid, glutamic acid, proline, isoleucine, leucine, tyrosine, lysine, histidine, arginine and total amino acids decreased, while the remaining threonine, serine, glycine, alanine, valine, methionine, phenylalanine remained relative stable. Besides aspartic acid, glycine, alanine, leucine, tyrosine and phenylalanine, other amino acids were all found to be significant different.⑥Among11volatile components, the content of isobutanol, isoamylol, nonanal and benzaldehyde decreased, while the content of glycerol increased. The changes of others were not obvious. Significant different was detected for all volatile parameters.The results indicated that most of routine parameters, lactic acid, elements, and amino acids, were involved in the complicated physical and chemical reactions in rice wine, and went through significant changes during aging. However, glucose was detected not to be significant different, which suggested that sugars might not go through obvious changes. The changes of most volatile compounds were not obvious, and further research was needed.(2) Aging status discrimination models were established based on the above chemical compositions. The results showed:①The accuracy obtained by SLDA1(base on PCs), SLDA2(based on chemical compositions) and PLS-DA models was50.6%,48.7%and 75.7%, respectively based on six routine parameters.②The accuracy obtained by SLDA1, SLDA2and PLS-DA models was39.3%,36.4%and66.5%, respectively based on sugars.③Based on nine mineral elements, the accuracy obtained by SLDA1, SLDA2and PLS-DA models was72.7%,73.4%and81.4%, respectively.④The accuracy obtained by SLDA1, SLDA2and PLS-DA models was91.0%,86.6%and91.2%, respectively based on routine parameters, lactic acid, sugars and elements.⑤SLDA1, SLDA2and PLS-DA models based on16amino acids gave the accuracy of86.4%,85.6%and91.2%, respectively.⑥The accuracy obtained by SLDA1, SLDA2and PLS-DA models was90.6%,90.6%and92.1%, respectively based on11volatile compounds.The results indicated that the discrimination models based on routine parameters, lactic acid, sugars and elements, as well as amino acids and volatile compounds reached a relative high accuracy over90%. These flavor components could be applied as fingerprint for aging status characterization of rice wine.(3) NIR spectra of base and blended rice wines in the range of800-2500nm were analyzed and compared. The main absorption peaks were all found to be at round1455nm,1692nm,1776nm,2266nm,2302nm and1900-1950nm, related to C-H, O-H, N-H functional groups in sugars, acids, amino acids of rice wines. PCA results showed that both base and blended rice wines of different aging status could be separated clearly. LDA, LS-SVM and PLS-DA were used to establish models for base rice wine age discrimination. The accuracy obtained were93.6%,98.7%and94.4%, respectively in calibration, and96.3%,91.3%and95.2%in prediction. For blended rice wine discrimination, the accuracy obtained in calibration were85.0%,100%,99.2%and100%by LDA, LS-SVM, PLS-DA and DPLS. And were72.5%,87.5%,86.3%and95.0%in prediction.The results indicated that NIR spectroscopy can obtain the holistic difference between samples, and can be used to establish fingerprint of rice wine of different aging status or ages.(4) PLSR quantification of routine parameters, lactic acid, sugars and amino acids in rice wine based on NIR and MIR spectroscopy was carried out. The results showed:①For the determination of routine parameters and lactic acid, both NIR and MIR spectroscopy could be used for quantitative analysis of total sugar, non-sugar solid and amino acid nitrogen. The RPD values of the models were all above2.0. Total acid, pH and lactic acid could be qualitative analyzed with RPD values above1.5. Besides lactic acid, the performances of NIR models were superior to MIR models.②For the determination of sugars, the quantification of glucose was satisfactory by both NIR and MIR spectroscopy with an RPD value above2.0. NIR was more suitable for analysis of isomaltotriose (RPD=1.95), while MIR was more suitable for analysis of maltose (RPD=1.97). For other parameters, the performance of the models was not satisfactory.③For the determination of amino acids, the performance of NIR models were generally much superior. Besides proline, phenylalanine, histidine and total amino acids, the RPD values of NIR models were all over1.5. In addition, the RPD values obtained for aspartic acid, isoleucine, tyrosine, and arginine were over2.0. The robustness of most MIR models was not satisfactory. RPD values obtained for aspartic acid, glutamic acid, isoleucine, phenylalanine and serine were above or close to1.5. For other amino acids, the robustness of the models was low (RPD<1.5).The results indicated that both NIR and MIR spectroscopy could be used for the analysis of routine parameters, lactic acid, sugars and some amino acids. But the performances of NIR models were generally better than MIR models.(5) The VIS/NIR spectra of rice wine samples in the range of303-1700nm under different temperature levels (5℃,10℃,15℃,20℃,25℃,30℃,35℃) were investigated. In high frequency (short-wave) region of1340-1480nm of raw spectra, the absorption increased when the temperature raised. On the contrary, in low frequency region, the absorption decreased when the temperature dropped. In second derivative spectra, the absorption at1320nm and1358nm increased with temperature. However, at1404nm, the absorption decreased with temperature. PCA results indicated that the samples under different temperature levels could be clearly separated. Because the hydrogen bonding will cause a shift of stretching vibration of water molecular to high frequency. As the temperature rises, it will promote hydrogen bonding, which might cause a change in the sample spectra.The determination of five routine parameters in rice wine was carried out by SMLR method under different temperature levels. The optimal temperature for quantification of alcohol degree, total sugar, non-sugar solid, total acid and amino acid nitrogen was found to be15℃,20℃,20℃,15℃and20℃, respectively. The model developed for alcohol degree and total sugar were relative robustness with an RPD value over2.5. The total acid model was moderate with an RPD value over1.5. However, the determination of non-sugar solid and amino acid nitrogen was not satisfactory due to low a RPD value (RPD<1.5). Temperature compensation models were established for alcohol degree, total sugar and total acid based on the spectra obtained under seven temperature levels. The compensation for alcohol degree showed good performance with a RPD value of the model to3.81. However, the compensation for total sugar and total acid was not satisfactory and needed to be improved (RPD<1.5).Finally, a hardware and software system was developed for quality determination of rice wine based on VIS/NIR spectroscopy, with the use of Zeiss portable spectrometer, optical fiber and temperature-controlled cuvette holder. The performance of the detection device was validated by37additional samples under temperature level of15℃. The results showed that:①For the determination of alcohol degree, the correlation coefficient in prediction, RMSEP and RPD was found to be0.979,0.464%and3.78, respectively.②For total sugar parameter, the correlation coefficient in prediction, RMSEP and RPD was0.766,5.78g/L and1.50, respectively.③The correlation coefficient in prediction, RMSEP and RPD was0.630,0.664g/L and1.24, respectively for total acid. The results indicated that the device might be suitable for alcohol degree detection in rice wine. But the precision for total sugar and total acid determination needed improvement. |
PDF Full Text Request