Abstract:This paper proposes a method for identifying different varieties of rice in Songyuan.For 368 rice samples from five varieties of Dao Huaxiang，Xiao Gaoliang，Tongxi926，Jijing515 and Nongda521 from Songyuan.Using Fourier transform infrared spectroscopy with a wave number in the range of 12 000 cm-1-4 000 cm-1 to acquire spectral data and preprocesses the data in six ways.The analysis results showed that the first derivative combined with SG9 point smoothing was the best pretreatment method.And using the partial least squares discriminant partial least squares discrimination analysis（PLS-DA）method to establish a discriminant analysis model for the corrected samples，and verify the model with the verification set.The model is validated by the combination of Dao Huaxiang，Xiao Gaoliang，Tongxi926，Jijing515 and Nongda521.The recognition rate of each variety was 100 %.And better than the results of principal component analysis.Dao Huaxiang samples from Liuhe and Meihe，and the same variety from Songyuan were used for discrimination.The results showed that the model can distinguish between Songyuan samples and non-Songyuan samples.This study provided a reference method for the identification and source identification of Songyuan rice varieties.