Analysis of Sugar Content of Wax Apple Fruit by Near Infrared Spectroscopy

• Main Author: 陳文誠
• Other Authors: 謝欽城
• Format: Others
• Language: zh-TW
• Published: 2000
• Online Access: http://ndltd.ncl.edu.tw/handle/75673025106529488608

碩士 === 國立屏東科技大學 === 機械工程系 === 88 === In this study, the near infrared (NIR) spectroscopic method was used to detect the sugar content of wax apple in Pingtung area by the 1100nm ~ 2200nm range. Total 596 wax apples were chosen to establish the calibration curves, and another 59 fruits were random were by sampled for the validation purpose. Both multiple linear regression (MLR) and partial least square regression (PLSR) were used in this study to develop the calibration models and to compare their differences. Five groups of wax apples from different locations in Pingtung in different seasons were used for five calibration sets and five prediction sets in the experiments. Five established calibration sets were merged as a total calibration set. The results of prediction from the total calibration curves were better than individuals. For all of the prediction sets, the results of prediction from the total calibration curves by PLSR were rp2 =0.800、0.626、0.851、0.787、0.779、0.847, and SEP=0.613、0.816、0.477、0.824、0.793、0.462 respectively. The results showed that the PLSR analysis was better than MLR analysis. As the factors =10, the results of calibration and prediction by PLSR were rc2 = 0.900, SEC = 0.480, rp2 = 0.847, SEP = 0.462. In MLR analysis, the best equation of calibration was obtained with ten-wavelengths, the results of calibration and prediction were rc2 = 0.824, SEC = 0.654, rp2 = 0.712, SEP = 0.639. For the math treatment, the best equation of calibration was obtained by second derivative spectra, therefore both MLR or PLSR used second derivative spectra to conduct the analysis of NIR.