Particulate matters (PM) have substantial influences on environmental system, climate change and public health. Ground based PM2.5 concentration measurement is insufficient in many circumstances. In this study, we using satellite retrieved AOD and other meteorological parameters such as the planetary boundary layer height (PBLH), temperature (TEMP), relative humidity (RH), U wind component (U), V wind component (V), surface pressure (SP), and large-scale precipitation (LSP), to establish GA-BP ANN AOD-PM2.5 retrieve model. The test R reached 0.83. This model is seasonally and regionally stable. The satellite AOD and ANN retrieved PM2.5 has the similar trend and distribution, and the trained model have practical as well as theoretical value.