Novel wind speed ensemble forecasting system based on the critic weighing principle of fuzzy information granulation and reverse mixed-frequency modeling

Accurate wind speed forecasting is essential for efficient wind power generation. It enhances wind turbine performance, increases power generation efficiency, reduces downtime, and minimizes operational and maintenance costs. However, most current wind speed forecasts are based solely on historic data, ignoring the potential of mixed-frequency and historical wind speed data, which results in poor projections. This work develops an ensemble forecasting system based on fuzzy information granulation and reverse mixed-frequency data modeling to maximize the potential of mixed-frequency and historical data. The system consists of four main modules: data noise reduction, the criteria importance through intercriteria correlation (CRITIC) principle of fuzzy information granulation, model library forecasting, and ensemble modules. Firstly, the noise reduction approach is applied to the original data to improve its quality. Secondly, the fuzzy information granulation methodology is utilized in conjunction with the CRITIC weighting principle to efficiently fuzzy granulate historical wind speed data in order to deal with data uncertainty. The reverse mixed-frequency sampling model, machine learning model, and statistical forecasting model are then employed as submodels for forecasting future wind speed changes. To make the most of a single model, we use the extreme learning machine (ELM) optimized by the ivy algorithm for ensemble forecasting. Finally, an empirical study is conducted using two data sets. The analytical findings reveal that the built forecasting system is accurate and stable. The system’s mean absolute percentage error (MAPE) values are 2.1197% and 2.3710%, respectively. Therefore, the proposed wind speed forecasting system is a very effective and dependable instrument for forecasting, which may offer substantial assistance in decision-making within relevant businesses.