Study on combination weight PLS model for determination of SSC in peach using the optimal informative regions obtained by different iPLS methods
基于iPLS原理最优化信息区间的桃糖度组合权重PLS模型研究

Backward interval partial least squares (BiPLS) and synergy interval partial least squares (SiPLS) are proposed to search for an optimized combination of informative spectral intervals about soluble solids content (SSC) from Vis/NIR spectra of peach. A linear combination weight PLS model was developed with all the separate PLS models in each regions. The spectra were preprocessed by second-order derivative and Savitzky-Golay smoothing. BiPLS procedure was used to search a combination of 742~770 and 862~920nm, and SiPLS method was used to locate a combination of 742~770, 832~860 and 892~920nm. The corresponding PLS models were built directly using the combination intervals, with the root mean square error of prediction (RMSEP) of 0.386 and 0.308, respectively. And two linear combinations weight PLS model were also constructed with the RMSEP of 0.351 and 0.364, respectively. The results reveal that the proposed methods are very promising for vibrational spectroscopy based on multicomponent analyses and give much better prediction than the whole-spectrum PLS modeling.