A novel weighted Root Mean Square Deviation (RMSD) control model based on the equally weighted sum of errors of the maximum, mean, and minimum values in each forecasting time period has been developed and applied with Holt-Winters forecasting method to do days or weekly ahead forecasting of daily averaged solar irradiance, output power, and energy efficiency for grid connected solar photovoltaic (PV) systems. The forecasting results based on the present model are compared favorably with the measured data of a grid connected solar PV system in Macau for solar irradiance, output power, and the system energy efficiency. Comparisons of the predicted results based on the present weighted RMSD control model to those from previous absolute error and relative error control models show that present weighted RMSD control model has better stability and higher accuracy. This thesis also develops new real time prediction models for output power and energy efficiency of solar photovoltaic (PV) systems. These models were validated using measured data of a grid-connected solar PV system in Macau. Both yearly and monthly averaged time frames are considered. It is shown that the prediction model for the yearly/monthly average of the minutely output power fits the measured data very well with high value of R square. The online prediction model for system efficiency is based on the ratio of the predicted output power to the predicted solar irradiance. This ratio model is shown to be able to fit the intermediate phase very well but not accurate for the growth and decay phases where the system efficiency is near zero. However, it can still serve as a useful purpose for practitioners as most PV systems work in the most efficient manner over this period. It is shown that the maximum monthly average minutely efficiency varies over a small range of 10.81%-12.63% in different months with slightly higher efficiency in winter months.