Comparison of Chemometrics Methods for Assessing Internal Quality of Pears On-Line using Visible/Near Infrared Transmission Technique

摘要

The objective of this research was to compare different chemometrics methods to assess internal quality of pears such as soluble solid content (SSC) and acidity (pH) on-line. Transmission spectra were collected by a fiber spectrometer in the wavelength range of 345-1040 nm, and the moving speed offruits was 0.5 m s~(-1). Spectral pre-processing including derivative, standard normal variate transformation (SNV), multiplicative scatter correction (MSC) was used before calibration. Chemometrics methodssuch as stepwise multiple linear regression (SMLR), principal component regression (PCR), partial least squares (PLS), artificial neural networks (ANN) and least squares support vector machines (LS-SVM) were compared. The results show that Vis/NIR spectroscopy is feasible for measuring SSC and pH value of pears on-line, PLS and LS-SVM methods are superior to other methods for assessing internal quality of pears.