Method for short-term photovoltaic generation power prediction base on weather patterns

This paper proposes a practical and effective method base on weather patterns for producing 15-minutes short-term prediction of photovoltaic generation with a prediction horizon of 24 hours. Two different radiation prediction models are examined for two different weather patterns. In the clear sky model, the surface radiation is predicted by calculating the cutting plane instantaneous intensity of solar radiation and atmospheric transparency coefficient. In the mode of non-clear sky, future radiation is forecasted by finding the quadratic curve of the measured surface radiation and the time. Finally, the radiation - power conversion model is developed to calculate the corresponding power prediction through the outputting of the short-term radiation prediction model. The proposed approach been applied in many PV power plant in the Northwest part of China. It is shown that the system can produce reasonable forecast results, providing a good guidance for dispatching and efficient running for PV power plant.