Abstract:

The objective of this study was to establish a comprehensive grading system for the quality evaluation of fresh
apple by online non-destructive testing. Factor analysis using SPSS 19.0 software was conducted to select the primary
indicators from 13 quality indicators of fresh Fuji apple from Luochuan, Shaanxi. Probability distribution and analytical
hierarchy process (AHP) were used respectively to grade the quality indicators and determine their weights. By employing
K-means clustering and discriminant analysis, the discriminant model of apple quality was established. The results showed
that the seven key quality indicators of fresh Fuji apple were single fruit weight, fruit shape index, color and lustre, soluble
solids, hardness, vitamin C and sugar/acid ratio. This study has established quality grading and evaluation standards, and
the five discirminant functions developed in this study could be applied to discriminate the quality of fresh apple with an
accuracy of 95.65% and 91.67% using modeling and test samples, respectively. This study can provide a scientific basis to
establish comprehensive assessment and grading standards for fresh apple quality, and also can offer technical support for
integrated on-line quality detection.

Key words: apple quality, factor analysis, analytical hierarchy process (AHP), cluster analysis, discriminant analysis, evaluation system