Construction Of Phytic Acid,Inorganic Phosphorus And Total Phosphorus Content Models Based On Visible And Near Infrared Spectroscopy In Rice

Phytic acid is an anti-nutritional factor and an important indicator of the nutritional quality of cereals.Phosphorus,one of the essential nutrients for the growth and development of animals and plants,plays an important role in a variety of physiological activities.This study aimed to construct calibration models using visible and near-infrared spectroscopy(Vis/NIRS)combined with chemometrics.The models constructed have achieved good prediction accuracy.The main results are as follows:1.The ferric chloride colorimetric method was used to determine phytic acid content and the molybdenum blue colorimetry was applied to measure inorganic phosphorus and total phosphorus contents of samples.The distribution range of phytic acid,inorganic phosphorus and total phosphorus contents in the samples were 0.0620?0.7624%,0.0165?0.0528%,0.0892?0.2510%,respectively.2.Correlations among phytic acid,inorganic phosphorus and total phosphorus contents in samples were significantly positive at 0.01 level,where,phytic acid content is highly positively correlated with total phosphorus content with the correlation coefficient r = 0.8570;the correlation between the phytic acid and inorganic phosphorus contents was moderately positive with r = 0.6285;the total phosphorus and inorganic phosphorus contents also performed moderately positive correlation with r = 0.6045.3.The range of phytic acid,inorganic phosphorus and total phosphorus contents in the calibration set can cover the range of corresponding components in the validation set very well,and the spectral information of the calibration set and the validation set overlap each other,indicating that the division of the calibration and validation set in this study is reasonable.4.The raw spectra of the samples were processed by eight spectral preprocessing methods such as smoothing,derivative and standard normal variate transformation(SNV),the best spectral pretreatment method was obtained by partial least squares(PLS).The optimal pretreatment methods for the spectra of phytic acid,inorganic phosphorus and total phosphorus models were moving average smoothing,savitzky golay smoothing and savitzky golay first derivative,respectively.5.Sensitive wavelengths for phytic acid,inorganic phosphorus and total phosphorus in the samples were obtained by competitive adaptive weighting algorithm(CARS)and correlation coefficient methods.There were no significant differences between models based on the fullwavelengths and models based on the characteristic wavelengths,indicating that the characteristic wavelengths selected for the above three components are effective.6.In the current research,the optimal calibration models for phytic acid,inorganic phosphorus and total phosphorus were the radio basis function least squares support vector regression(RBF-LSSVR)model based on the characteristic wavelengths using polished rice flour samples(Rp2-0.9541,RMSEP = 0.0298),the PLS model based on the sensitive wavelengths of using polished rice flour samples(Rp2 = 0.9082,RMSEP = 0.0022),and the linear least squares support vector regression(Lin-LSSVR)model based on the full wavelengths using polished rice flour samples(Rp2= 0.9541,RMSEP = 0.0060),respectively.The results showed these models own powerful prediction performance.