| Rice is one of the most widely cultivated food crops in the world.With the progress of society,the market share of high-quality rice is getting higher and higher.Therefore,it is one of the important tasks of rice breeding to breed high quality and high yield rice varieties.Traditional rice quality chemical determination method is difficult to meet the demand of rapid quality detection of large quantity rice varieties(combinations)due to its complex pretreatment,complicated process and high consumption of reagents.Nir spectroscopy can quickly,harmlessly and pollution-free detect the quality of rice products.The study on NIR detection of amylose content and gum consistency of rice is helpful for early breeding of high-quality rice varieties(combinations).There are abundant rice germplasm resources in southern rice area.137 varieties(combinations)of different main rice varieties were collected in the experiment.The content and consistency of amylopectin in rice were detected and analyzed by chemical method,and the NIR spectra were collected.To provide support for the rapid detection and screening of early generations of high-quality rice varieties.The main results of this study are as follows:1.Nir Modeling of amylopectin content in rice:16 mathematical pretreatments and 4wavelength bands(or combinations)were used to establish the NIR analysis models of amylopectin content and NIR spectral values of rice materials by the least square method.According to the model evaluation indexes,the smooth pretreatment was determined as the best pretreatment method,and the wavelength segment from 1100 nm to 1650 nm was the best modeling wavelength segment.The calibration correlation coefficient(R),test correlation coefficient(r)and relative analysis error(RPD)of the evaluation index of the smoothing processing model were respectively 0.9494,0.9427 and 3.0598.The calibration correlation coefficient(R),test correlation coefficient(r)and relative analysis error(RPD)of the 1100 nm to 1650 nm model in the wavelength range were 0.9529,0.9445 and 3.1090,respectively.After eliminating two abnormal samples under the optimal modeling condition,an optimal rice amylose content NIR data analysis model was obtained,and the calibration correlation coefficient(R),test correlation coefficient(r)and relative analysis error(RPD)were respectively 0.9565,0.9495 and 3.2673.Then 30 validation sets were used for external validation of the model,and paired T-test was performed for the predicted value of the model and the chemical test value.The results showed that the absolute error of predictive value and chemical value of the model ranged from 0.0135% to 1.0811%,and the P value of paired T-test was 0.934 greater than 0.05.The model can be used to detect amylose content in rice.2.Nir Modeling of rice gum consistency: The NIR and NIR spectral values of rice materials were respectively constructed by using 20 mathematical pretreatments and 6wavelength segments combined with partial least squares(PLS).The results show that the modeling effect of smooth pretreatment is the best,and the calibration correlation coefficient(R),test correlation coefficient(r)and relative analysis error(RPD)are respectively 0.9700,0.9642 and 3.7805.The parameters of the wavelength range from 1100 nm to 1650 nm were the best,R,r and RPD were 0.9694,0.9638 and 3.7586,respectively.The optimal near infrared analysis model of rice glue consistency was established in the smoothing treatment and the optimal wavelength section.The R,r and RPD were 0.9790,0.9741 and 4.4194,respectively.Then,the optimal model was externally verified and the paired T-test was conducted for the model detection value and the original chemical value.The absolute error of the result was 0.1986 mm-6.5024 mm,the paired T-test P = 0.726 was greater than 0.05,and there was no significant difference between the NIR spectral value and the chemical value,indicating that the NIR analysis model could be used to detect the consistency of rice glue. |
PDF Full Text Request