| Corn is one of the three major food crops in China,with an annual output of 500 million tons,which is of great significance to national food security.During the storage process,the quality of corn will gradually deteriorate with the change of storage time,and the deterioration rate is easily affected by its initial moisture content and storage temperature.In the process of corn storage,fatty acid values need to be measured to determine the state of corn to better adjust the storage strategy of corn.However,the traditional method to determine the fatty acid value of corn is time-consuming and labor-intensive,and can not achieve the ideal effect.In this paper,the quality change of corn with different initial moisture content under different storage temperatures was studied,and the characteristic index of quality change of corn during storage-fatty acid value was selected by PCA and PLS-DA.Near Infrared Spectroscopy(NIR)and Low Field NMR(LF-NMR)combined with chemometrics were used to establish a quantitative detection model of corn fatty acid values.The quality variation of maize during storage was studied.The results showed that with the increase of storage time,storage temperature and water content,fatty acid value,minimum viscosity and final viscosity gradually increased,catalase activity,decay value,T21 and P21 gradually decreased,while corn T22,P22,T23 and P23 did not change significantly during storage.The indexes with VIP value greater than 1 were selected by PCA and PLS-DA,including fatty acid value,catalase activity,minimum viscosity and final viscosity,among which the VIP value of fatty acid value was the largest.Therefore,fatty acid value was selected as the sensitive index of quality deterioration during storage of corn.The quantitative model of fatty acid value during corn storage was constructed by near infrared spectroscopy.Firstly,a variety of spectral pretreatment methods were used,and it was found that the standard normal variable change(SNV)method was better.Then competitive adaptive reweighting algorithm(CARS)and joint interval partial least squares(Si)are used to obtain the characteristic wavelength,and partial least squares(PLS)quantitative prediction models based on full spectrum and characteristic spectrum are established respectively.The RMSEP of NIR-PLS,CARSPLS and Si-PLS were 0.672,0.984 and 0.705,respectively.The prediction accuracy of CARS-PLS model is higher than that of NIR-PLS and Si-PLS.A quantitative model of corn fatty acid value was constructed by using near infrared spectroscopy and low field nuclear magnetic attenuation curve,and a fusion model of fatty acid value prediction was established by fusing NIR and LF-NMR data of corn samples with feature layer and data layer.The model results showed that the BP-NN model based on the feature layer data had the best performance in the prediction of corn fatty acid value,and the correlation coefficients Rc and Rp of the correction set and prediction set were as high as 0.9997 and 0.9698.The accuracy and stability of the model are higher than that of the prediction using data layer and single technique,and the feasibility of the fusion of NIR and LF-NMR techniques in the prediction of fatty acid values during corn storage is verified. |
PDF Full Text Request