A Fast Identification Method Of Oilseed Crops Quality Based On Low Field Nuclear Magnetic Resonance Technology

Soybean and sunflower-seed are important oilseed crops in China,which are popular because of their rich nutrition.Currently,there are a great variety of oilseed crops in the market,and good and bad products are mixed up.In particular,bad products damage the profits of farmers.Therefore,it is urgently needed to establish methods for monitoring the oilseed crops quality.As the first part of this thesis,various varieties of sunflower-seeds were discriminated and the oil contents of these sunflower-seeds were determined by low field-nuclear magnetic resonance(LF-NMR)technology combined with chemometrics,which was conducted to establish a fast identification method of sunflower-seed quality.The Carr-Purcell-Meiboom-Gill(CPMG)relaxation delay data were analyzed by principal component analysis(PCA).It was found in the PCA scores plot that 368,3638 and 610 sunflower-seeds were located in different areas,identifying an effective variety discrimination of sunflower-seeds by PCA method.Partial least squares regression(PLSR)models for the oil determination of sunflower-seeds were created by using the oil contents obtained by Soxhlet extraction method as the measured values and the LF-NMR data as the predicted values.The results indicated that the models were satisfactory with high Rcal2(0.9705)and low RMSEC(0.6960%).Furthermore,the oil contents in unknown sunflower-seed samples were predicted by the PLSR model and compared with the reference values with the relative error in the range of 1.06%-7.36%.It's noteworthy that the model showed the best prediction accuracy for the 610 sunflower-seeds with the relative error from 1.06%to 2.34%.Therefore,the combined method of LF-NMR and chemometrics can identify the sunflower seed quality fast and noninvasively.Furthermore,low field nuclear magnetic resonance technology combined with chemometric method was employed to simultaneously determine the oil and water contents in soybean.The CPMG magnetization decay data of Jilin soybean samples were acquired by LF-NMR and directly applied for PLSR.Calibration and validation models were created by PLSR with full cross-validation on the data obtained by Soxhlet extraction method for oil and oven-drying method for water as references.The results indicated that the calibration models were satisfactory for both oil(Rcal2>0.97)and water(Rcal2>0.99);the root mean squared errors of cross-validation(RMSECV)for oil and water were 0.2722%and 0.2902%by PLSR.Furthermore,the oil and water in unknown soybean samples were predicted by the PLSR model and compared with the reference values,with the relative errors in the range of 1,25%-4.96%for oil and 0.44%-2.49%for water.These results showed that the combination of LF-NMR relaxometry with chemometrics can be applied for the determination of oil and water contents of soybean with high accuracy.Finally,LF-NMR and MRI techniques were introduced to monitor a typical fermentation process of natto.Upon a multi-exponential fitting for the NMR data obtained by a CPMG pulse sequence,the spin-spin relaxation time(T2)spectra displayed the presence of four populations of water molecules during the fermentation:bound water(T21),immobilized water(T22),moderate immobilized water(T23)and free water(T24).The concomitant increases in apparent physicochemical properties and the T23 water populations could be correlated with the extended fermentation time.The chemometric method of PCA based on the CPMG data further revealed that the fermentation is characteristic of a process with three main phases.On the other hand,the proton density weighted images,derived from MRI,revealed the spatial distribution pf water molecules within the fermented products,and projected a full hydration process inside natto through a gradual penetration of water molecules from the edge to the center of the granules.Therefore,LF-NMR and MRI combined with chemometrics provide unique insights into the monitoring of natto fermentation.