Abstract: In order to explore the applicability and accuracy of PM_2.5 concentration prediction method in atmospheric environment,this paper established linear regression and BP neural network prediction models,and compared and analyzed the prediction effect of the two models on PM_2.5concentration based on the correlation between PM_2.5 concentration in atmospheric environment and meteorological factors such as temperature and rainfall.The results showed that the PM_2.5 concentration in winter was higher than that in summer,indicating a trend that the lower the atmospheric temperature,the higher the concentration level.When the daily rainfall exceeds a certain threshold,it is conducive to the dilution of PM_2.5 concentration.There were differences in the prediction results between linear regression and BP neural network model.The relative errors of linear regression model were both lower than 30%.The overall prediction effect was better than that of neural network model.When linear regression prediction model was used with a small amount of sample data,the reliability and accuracy of prediction results became better.