Based on the aerosol optical inversion data from AERONET SGP station, an aerosol classification model based on a naive Bayesian classifier is proposed. The single scattering albedo and complex refractive index of aerosol are used as input variables to identify five types of aerosols in this region, and the optical properties of different types of aerosols are analyzed. The proposed model generates a classifier model based on the classification probability distribution of the training sample sets, and then predicts the classification of the test sample sets. On this basis, the proposed model is used to analyze the seasonal distribution difference characteristics of the aerosol types at the SGP station, and the experimental results are consistent with the climatic environment characteristics of this region. In order to verify the accuracy of classification results of the proposed model, the aerosol classification threshold standards are established by using the matching method combining AERONET station data and high spectral resolution lidar profile data. The results show that compared with the traditional aerosol classification algorithm, the aerosol classification results obtained by the proposed model have a high consistency with the results determined based on threshold criteria, which can provide ground data support for aerosol inversion by remote sensing equipment such as satellites.