Wind speed forecasting by neural network constructed for each wind direction

摘要

In order to reduce the forecast error of the mean wind speed forecasting, the forecasting method in which the forecast models are constructed for each wind direction is proposed. First, the forecasting models by multiple regression analysis (MRA) are constructed for each wind direction, and suitable data input period and interval are examined. Next, the forecasting models by neural network are constructed for each wind direction, and wind speeds are forecasted by inputting data at suitable data input period and interval to those models. As compared with the conventional forecasting method, a forecasting error is reduced 2 to 7 point by the proposed method.