Abstract
This paper reports an insolation forecasting method for photovoltaic power predictions. The method of proposal forecasts the global irradiance every one hour by using weather forecast every three hours. The weather forecast is classified into 14 kinds of weather change categories by weather change patterns, and the insolation forecasting accuracy is improved. The forecasting accuracy was examined based on the measurement data and the weather forecast announced in Tokyo. In the result, using weather change patterns decrease mean absolute error ratio of hourly forecasting from 32.6 to 30.2% and the error ratio of daily forecasting from 24.7 to 23.5%. The result also shows the possibility that the error ratio of hourly forecasting is decreased to 24.5% and the error ratio of daily forecasting is decreased to 17.8% in Tokyo when the weather forecast accuracy is improved.
