| China accounts for the major chestnut production in the world.The accurately detection of chestnut quality gives great significance to increasing export profits.The quality of chestnut is mainly related to its sweetness and waxy index.The sweetness of chestnut is greatly affected by the content of moisture,water-soluble sugar and sucrose.The total starch content and amylose content of chestnut can make a rough judgment for the waxy index,and an intuitive evaluation can be made based on its texture characteristics.At present,the commonly used measurement method for the main chemical components of chestnut is chemical analysis,which is time-consuming and loss in edible value after measuring the samples.Near-infrared?NIR?diffuse reflectance spectroscopy is a fast,non-destructive and accurate detection technology that has been widely used in the food science.In the present paper,NIR diffuse reflectance spectroscopy combined with chemometrics and fuzzy mathematics methods were used for the rapid determination of chestnut sweetness indices and waxy value,which provides a basis for the quality control of chestnut.The main results were as follows:?1?For the rapid determination of chestnut sweetness,the PLS model pretreated with the first derivative+Savitzky-Golay?5 points smoothing?in the 5900-5300 cm-1 band was the optimal model of moisture content.The R2 of calibration model was 0.9639,the R2 of verification model was 0.9071,the corresponding values of RMSEC,RMSEP and RMSECV were 0.0121,0.0186 and 0.0316 respectively.The PLS model combined with the first derivative+Savitzky-Golay?5 points smoothing?at the full-wave band was the optimal model of water-soluble sugar content,the R2 of calibration model was 0.9257,the R2 of verification model was 0.8719,the corresponding RMSEC,RMSEP and RMSECV values were 0.0162,0.0217 and 0.0299,respectively.The optimal model of sucrose content was obtained by PLS method combined with the first derivative+ Norris?segment length 1,interval 1?preprocessing at the full-wave band.The R2 of calibration model was 0.9999,the R2 of verification model was 0.9737,and the corresponding RMSEC value,RMSEP and RMSECV values were 0.0251,0.435 and 0.567,respectively.?2?For the rapid determination of starch content and amylose content in chestnut,the optimal model of starch content was obtained by PLS method using the first derivative combined with Norris?segment length 3,interval 1?.The R2 of calibration model was 0.9671,the R2 of verification model was 0.9537,the corresponding RMSEC value,RMSEP and RMSECV values were 0.639,0.802 and 2.07,respectively,and the optimal model of amylose content was achieved in PLS models at 7200-4000 cm-1,with the first derivative,the first derivative+Savitzky-Golay?3 points smoothing?and the first derivative+Norris?segment length 1,interval 1?as the best pretrentment methods.The R2 of calibration model was 0.9984,the R2 of verification model was 0.9008,and the corresponding RMSEC,RMSEP,and RMSECV values were 0.0846,0.558,and 1.66,respectively.?3?For the rapid determination of the glutinous value of chestnut,fuzzy mathematics,sensory evaluation and MATLAB software were successfully used in calculating the value of glutinous index of chestnut.The model of the glutinous value of chestnut constructed by near-infrared diffuse reflectance spectroscopy achieves the best modeling effect when using the first derivative+Savitzky-Golay?5-point smoothing?in the 9400-4000 cm-1 band.The R2 of calibration model was 0.9805,the R2 of verification model was 0.9127,and the RMSEC,RMSEP.and RMSECV values were 0.246,0.979,and 2.440,respectively.In summary,the PLS models of sweetness indices,starch,amylose and glutinous value of chestnut were establish by the first derivative combined with smoothing pretreatment,which achieved accurate and robust prediction results.Therefore,near-infrared spectroscopy combined with chemometrics and fuzzy mathematics can quickly and accurately determine the quality of chestnut. |
PDF Full Text Request