Use of FT-NIR spectrometry in non-invasive measurements of soluble solid contents (SSC) of 'Fuji' apple based on different PLS models

作者：Zou Xiaobo^*; Zhao Jiewen; Huang Xingyi; Li Yanxiao

来源：Chemometrics and Intelligent Laboratory Systems, 2007, 87(1): 43-51.

DOI：10.1016/j.chemolab.2006.09.003

• apple
• soluble solid content (SSC)
• near infrared
• variables selection
• PLS

The feasibility of using the efficient selection wavelength regions in FT-NIR for a rapid and conclusive determination of fruit inner qualities such as soluble solid content (SSC) of apples was investigated. An apple's MRS acquisition device was developed in this study. With this device, the apple was rolling while collecting the NIR spectroscopy. Graphical ly-oriented local multivariate calibration modeling procedures called interval partial least-squares (iPLS), backward interval partial least-squares (BiPLS), forward interval partial least-squares (FiPLS), genetic algorithm interval partial least-squares (GA-iPLS) and genetic algorithm partial least-squares (GA-PLS) were applied to select the efficient spectral regions that provide the lowest prediction error, in comparison to the full-spectrum model. The optimal combinations of 12 spectral intervals among 40 intervals which selected by BiPLS yielded a good result (RMSEC=0.39286, r(c)=0.953, RMSEP=0.4531, r(p)=0.924), and the optimal combinations of 14 spectral intervals among 40 intervals which selected by FiPLS also yielded good results (RMSEC=0.4027, r(c)=0.948, RMSEP=0.4424, r(p)=0.927). The intervals selected by BiPLS were not the same with. those selected by FiPLS, due to different algorithms employed in these two methods. The five intervals selected by GA-iPLS which contained 360 variables were used as calibration set in GA-PLS model. 45 variables were selected in the best prediction ability of the GA-PLS models (RMSEP=0.414). These procedures allow the simultaneous determination of other internal quality of fruits.

• 出版日期2007-5-15
• 单位江苏大学