Abstract: To accurately forecast the generator of grid\|connected PV power system and to improve its security and stability, the output power of a single component of silicon solar cell is calculated in this paper. In addition, a multivariate linear regression forecasting model is proposed with variables of radiation and temperature. A test scheme was designed firstly, then operation data obtained from the PV system in sunny day, cloudy day and the day from fine to cloudy or the day from overcast to cloudy is collected. Finally, the regression equations for three kinds of weather are solved based on forecasting model that is applied in 2kW grid\|connected PV system. The forecasting results show that the forecasting method is not only simple and has high accuracy, but also suitable to solve stochastic problems to improve the operation stability of PV system.