Determination Of Quality Parameters In Intact Rapeseed By Portable NIRS Analyzer

Near infrared reflectance spectroscopy technology and 322 rapeseeds selected samples were used to set up models for determination of seven quality parameters in intact rapeseeds.Cross validation and 320 test samples determination showed that NIRS was a non-destructive,rapid,efficient and pollution-free technique.It was thus concluded that NIRS analysis can be applied to quantitative analysis of quality parameters of rapeseed.1.The wavelength region of erucic acid,oleic acid,linolic acid,oil content,protein content,total glucosinolate and water content optimized by chemometrics were 1300-1836 nm,1300-2356 nm,1300-2500 nm,1300-2500 nm,1300-2500 nm,1638-2356 nm and1300-2500 nm,respectively.2.The spectra preprocessing of the seven quality parameters were multiplicative scattering Correction+first derivative,normalization,multiplicative scattering Correction+first derivative,standardization+MSC,standardization+MSC,normalization,standardization+MSC,respectively.3.The correlation coefficient of PCR for erucic acid calibration and validation model were 0.90 and 0.80,the root mean square error of cross-validation variance(RMSECV) were 4.8 and 3.5;The correlation coefficient of PCR for oleic acid calibration and validation model were 0.92 and 0.85,RMSECV were 4.8 and 12.0;The correlation coefficient of PCR for linolic acid calibration and validation model were 0.95 and 0.75,RMSECV were 0.8 and 2.3;The correlation coefficient of PCR for oil content calibration and validation model were 0.95 and 0.93,RMSECV were 1.00 and 1.12;The correlation coefficient of PCR for protein content calibration and validation model were 0.95 and 0.83, RMSECV were 0.68 and 1.83;The correlation coefficient of PCR for total glucosinolate calibration and validation model were 0.85 and 0.76,RMSECV were 7.90 and 8.32;The correlation coefficient of PCR for water content calibration and validation model were 0.98 and 0.75,RMSECV were 0.56 and 0.38;4.Partial least square and principle component regression for setting up the models were used.The results showed that principle component regression was faster than partial least square and gave a higher coefficient correlation.The Near Infrared Reflectance Spectroscopy Technology was predicted to have the"potential of solving the problems in the global agriculture analysis"by the global field of agriculture analysis for its merits of harmlessness,rapidity,environmental friendliness and portability.It was regarded as the direction of future development of the present technology for the double low rapeseed (low erucic acid and low glucosinolate) quality evaluation.