Abstract:A non-destructive method for rapidly measuring the sugar content in passion fruits was established.The near-infrared spectroscopy was combined with the synergy interval partial least squares（SiPLS）and the competitive adaptive reweighted sampling（CARS）algorithms to screen the characteristic wavelengths of the near-infrared spectroscopy.The partial least squares and support vector machine were used to establish the prediction model of sugar content in passion fruits.The results showed that the model established by multivariate linear regression outperformed that established by multivariate nonlinear regression，and the number of characteristic wavelengths screened by the SiPLS-CARS algorithms was 67，accounting for 2.90% of the full spectrum.The prediction model showed the R2c of 0.972 7，the root mean square error of calibration（RMSEC）of 0.333 8，the R2p of 0.967 2 and the root mean square error of prediction（RMSEP）of 0.366 0 on the validation set，and the relative prediction deviation（RPD）of 4.506 6.The established method can realize non-destructive and rapid measurement of passion fruit sugar content，and the model in the portable equipment for non-destructive measurement of passion fruit sugar can be simplified.