An improved multi-step forecasting model based on WRF ensembles and creative fuzzy systems for wind speed

Abstract

Accurate wind speed forecasting, which strongly influences the safe usage of wind resources, is still a critical issue and a huge challenge. At present, the single-valued deterministic NWP forecast is primarily adopted by wind farms; however, recent techniques cannot meet the actual needs of grid dispatch in many cases. This paper contributes to a new multi-step forecasting method for operational wind forecast, 96-steps of the next day, termed the CS-FS-WRF-E model, which is based on a Weather Research and Forecasting (WRF) ensemble forecast, a novel Fuzzy System, and a Cuckoo Search (CS) algorithm. First, the WRF ensemble, which considers three horizontal resolutions and four initial fields, using a 0.5° horizontal grid-spacing Global Forecast System (GFS) model output, is constructed as the basic forecasting results. Then, a novel fuzzy system, which can extract the features of these ensembles, is built under the concept of membership degrees. With the help of CS optimization, the final model is constructed using this evolutionary algorithm to adjust and correct the results obtained based on physical laws, yielding the best forecasting performance and outperforming individual ensemble members and all of the other models for comparison.