Premium
Prediction of Firmness and pH for “Golden Delicious” Apple Based on Elasticity Index from Modal Analysis
Author(s) -
Hou Jumin,
Zhang Yuxia,
Sun Yonghai,
Xu Na,
Leng Yue
Publication year - 2018
Publication title -
journal of food science
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.772
H-Index - 150
eISSN - 1750-3841
pISSN - 0022-1147
DOI - 10.1111/1750-3841.14071
Subject(s) - partial least squares regression , mathematics , modal , regression analysis , multivariate statistics , outlier , regression , test set , elasticity (physics) , biological system , statistics , chemistry , materials science , biology , composite material , polymer chemistry
An experimental modal test system was established to extract the natural frequencies of “Golden Delicious” apple, after which the elasticity index was calculated to predict the apple quality parameters based on the orthogonal polynomials method. The elasticity index in every vibration mode changed dramatically ( P = 0.01) along time revolution. The multivariate regression methods were used to model the predictive relationship between the elasticity index and the apple quality parameters. The models of the apple juice pH based on support vector regression presented adequate determination coefficients of calibration set ( Q 2 = 0.68) and prediction set ( R 2 = 0.55), respectively. The models based on partial least squares regression could be used for predicting the apple firmness parameter offset gradient ( Q 2 = 0.76 and R 2 = 0.72). It helped understanding the fruit dynamic properties of the fruit and spontaneously obtaining the fruit chemical parameters. A nondestructive and portable device was viable for fruit quality estimation by the modal test system during storage, transport, and even growth on the tree. Practical Application A nondestructive and portable device was provided for fruit quality detection during storage, transport and even growth based on experimental modal analysis. A systematic statistical analysis method about outlier detection, data set partitioning, parameter optimization, and multiple regression models were provided.