Combined with the abnormal data in the aerosol mass concentration observation of Guiyang Meteorological Station, the corresponding quality control measures are put forward. The hourly data checking program is compiled to remind the attendants to check the alarm reasons in time, so as to minimize the loss of original data and the generation of outliers. By using 7-5-3 Hanning smoothing filtering method, the processed PM2.5 data are compared with the original minute data.The results show that this method can eliminate the outliers while retaining the variation characteristics of the original sequence .Based on the quality control data of PM2.5 from 2013 to 2016 and the meteorological data of the same period, the variation characteristics of PM2.5 mass concentration are analyzed briefly.The results show that the PM2.5 monthly mean concentration presents a single valley and multi-peak trend,it was high in winter and low in summer.The correlation between PM2.5 mass concentration and meteorological factors in the same period was analyzed with an air pollution period of 9 consecutive days in January 2013 as an example. The data showed that PM2.5 was negatively correlated with wind speed and precipitation. That is, the higher the wind speed, the smaller the PM2.5 concentration, the more obvious the precipitation effect on purifying air, and the lower the PM2.5 concentration.