Prediction of watermelon quality based on vibration spectrum

Abstract

Judging watermelon quality based on its apparent properties such as size or skin color is difficult. A non-destructive method is employed here, based on vibrational response spectrum, to determine the quality indices of watermelon (Charleston gray). The responses of samples to vibration excitation were recorded by laser Doppler vibrometry (LDV). The phase shift between input and output signals were extracted over a wide frequency range. The total soluble solids (TSS), titratable acidity (TA) and TSS/TA ratio also measured as watermelon quality characters. Stepwise multiple linear regression (SMLR) as well as partial least square regression (PLS) was applied to extracted vibration spectrums to construct prediction models of watermelon quality. The results showed that performance of SMLR models were better than PLS. The determination coefficients (R2) of SMLR validation models were 0.9976, 0.9985 and 0.9542 for TSS, TA and TSS/TA respectively. It is likely that reduction of cell wall materials to soluble solids during ripening process changes viscoelastic properties of watermelon reflected by vibrational response. This study demonstrated the feasibility of mentioned method for predicting the quality of watermelons in an industrial grading system.