A Short-Term Ensemble Wind Speed Forecasting System for Wind Power Applications

Abstract

Accurate short-term wind speed forecasts for utility-scale wind farms will be crucial for the U.S. Department of Energy's (DOE) goal of providing 20% of total power from wind by 2030. For typical pitch-controlled wind turbines, power output varies as the cube of wind speed over a significant portion of the power output curve. Therefore, small improvements in windspeed forecasts would constitute much larger improvements in wind power forecasts. In addition, communicating the level of uncertainty in these wind speed forecasts will allow the industry to better quantify the level of financial risk inherent with these forecasts. In this study, a computationally efficient and accurate forecasting system is developed. This system uses a 21-member ensemble of the Weather Research and Forecasting Single-Column Model (WRF-SCM V3.1.1) to generate a probability distribution function (PDF) of 1-hour forecasts at a 90m height location in West/Central Illinois. The WRF-SCM ensemble was initialized by the 20 km Rapid update Cycle (RUC) 00h forecast and perturbed by both perturbations in the initial conditions and physics options. The PDF was calibrated using Bayesian Model Averaging (BMA) where the individual forecasts were weighted according to their performance. This combination of a mesoscale numerical weather prediction ensemble system and Bayesian statistics allowed for both accurate prediction of 1-hour wind speed forecasts and their level of uncertainty.iii

A vector autoregression weather model for electricity supply and demand modeling

“…More short-term solar radiation forecasting is done using cloud motion derived from satellite images [11,12,19]. Perez et al [12] report an increase in the RMSE% from 25% to 42% as the forecasting horizon goes from hour-ahead to six-hours-ahead.…”

“…Perez et al [12] report an increase in the RMSE% from 25% to 42% as the forecasting horizon goes from hour-ahead to six-hours-ahead. Traiteur et al [19] compare their forecasts against single point ground-truth stations and report RMSEs that vary from 68 Wh=m 2 to 120 Wh=m 2 for hour-ahead forecasts and 140 Wh=m 2 to 200 Wh=m 2 to six-hour-ahead forecasts. Erdem and Shi [17] generate one-, two-, and three-hours-ahead solar forecasts and report RMSE%s of 23%, 32%, and 38%, respectively.…”

“…Liu et al [18] propose a novel time-series technique that is based on the Taylor Kriging model. Other works combine multiple numerical techniques to produce ensemble wind forecasts [19][20][21]. Wang and Xiong [22] develop a hybrid forecasting method based on an ARMA process, outlier detection, and fuzzy time series to forecast the daily wind speed in Taiwan.…”

Daily Average Wind Power Interval Forecasts Based on an Optimal Adaptive-Network-Based Fuzzy Inference System and Singular Spectrum Analysis

“…Sustainability 2016, 8,125 22 of 31 models, and being more to the right on the IFCP axis indicates a higher IFCP. Thus, the IFASF model's IFCPs are higher than other models' IFCPs, demonstrating that SSA and FA actually improve the interval forecasting ability of ANFIS and show a good pre-processing and optimal effectiveness for randomly selected weeks.…”

A Long-Term Wind Speed Ensemble Forecasting System with Weather Adapted Correction

“…Therefore, improvements have been made by adding more parameters for better estimation accuracy. Previously, parameters to be added in statistical models mostly focus on day/night flag [16], the forecast length [17] and seasonality, as these parameters are physically related to prediction errors and are able to gain along with the prediction. On the other hand, predictions of other meteorological variables besides the wind speed are often added in models.…”