| Recently, with the development in agricultural product processing and agricultural modernization of our country, the technology of non-destructive inspection and classification of farm products is becoming more and more important and imperative.The internal quality of the fruits reveals their dielectric, chemical and physical property. The biochemical reaction during fruits' growing, ripeness, being damaged and being rotted away occurs with the invertion of material energy, which causes the change of chemical compositions in fruits. As a result, the quality of fruits will change. These fruits with internal defects can be taken out in advance by inspecting the internal quality of fruits. Furthermore, the processing technology of fruits will also be effectively improved if fruits could be classified strictly according to their quality before being sold or circulating.Traditionally, the quality of fruit was mainly determined according to sugar content, organic acid and the ratio of sugar to acid with chemical analytical methods. These methods are destructive, expensive, complex, time and labor consume and off-line and can not satisfy the practical demand for fast antomatic grading of fruits.The near infrared spectral analysis is a new nondestructive detecting technology, by which fruit compositions are determined according to absorbing, scattering, reflection and transmission of near infrared spectra. The near infrared spectroscopy (NIRS) has gained wide acceptance in different fields by virtue of its advantages such as fast, non-destructive and ability to record spectra for solid and liquid sample's with no prior manipulation since 90's.Pear is common and favorite fruit, which is well known by luscious, crisp and succulent. In this study, famous Dang shan pear were chosen as the respective research object. The objective of the present research was to study the potential of NIR diffuse reflectance spectroscopy as a method for non-destructive measurement of sugar content (SC), valid acidity, firmness and density of Dang shan pear.Fourier transform near infrared spectroscopy instrument was used to select spectra of Dang shan pear to relate SC, valid acidity, firmness and density to spectra. Different spectral pretreatments and selected spectral ranges were used to optimize the calibration models. Mathematic models used to predict pear's quality were established by two multivariate calibration techniques: partial least square (PLS) and principal component regression (PCR) methods using the optimization spectra pretreatment and spectral range. The models accuracy and prediction ability were compared and vertified. Experiment results showed that the better prediction models by PLS for sugar content (SC), valid acidity(VA), firmness of the sounds and SC of the bruises were established in two spectral regions: 9301～4532.55cm-1 and 9967～4030cm-1 respectively, and yielded correlation coefficient of calibration 0.940, 0.862, 0.865, 0.992, root mean square error of calibration (RMSEC) 0.292, 0.047, 0.191, 0.157 and root mean square error of prediction (RMSEP) 0.349, 0.081, 0.310, 0.854, respectively. Correlation coefficients of all models are above 0.85, indicating the feasibility of FT-NIR spectral analysis for predicting internal quality of Dang shan pear. |
PDF Full Text Request