Abstract:Based on the power generation data provided by the PV power station of Hubei Meteorological Energy Development Centre and the meteorological data of the corresponding period, we carried out verification and evaluation on the radiation and generating power short term forecast method and found that: (1) Good correlation exists between the solar irradiance forecasts and the actual situation, with correlation coefficient going above 0.77, passing the 0.001 significance test. (2) The optimal method of short term forecasting methods is putting mode irradiance into photoelectric conversion model, and the relative root mean square error for the first day is 0.16. (3) Solar radiation forecasts and PV power generation forecast with solar elevation angle change present certain regularity. The maximum error is found at winter noons and the minimum is over summer nights. The error is significantly higher in rainy days than in sunny days. How to reduce the error in the rainy weather forecast is the study focus in the next step work.