Correction Method for Wind Speed Forecast in Wind Farms Based on Kernel Density Estimation and Multiple Model Approaches

Wind power plays a significant role in the global capacity of renewable energy sources. It is essential to conduct research on wind power forecast to effectively plan for wind power generation and grid dispatch. Wind speed forecast serves as the fundamental basis for accurate wind power forecast. However, forecasted wind speed in wind farms often involves significant errors and does not easily conform to prior distribution functions. This issue significantly affects the precision of wind power forecast. To address this problem, this study proposes a wind speed correction method for wind farm forecasting that relies on non-parametric estimation and multi-models. In this paper, we analyze the distribution characteristics of wind speed forecasting errors in wind farms and utilize non-parametric estimation methods to accurately account for forecasting errors at various wind speed levels. Furthermore, we extract the characteristics obtained from the fitting results. By applying the FCM method to cluster and classify the error characteristics within different wind speed intervals, we establish and validate a multi-model framework for correcting wind speed in wind farm forecasting.