Precise wind speed prediction is increasingly practical for sustained and stable wind energy utilization considering the growing portion of wind energy in the global electric grid. Although plenty of wind speed forecasting approaches have been devoted to improving forecasting performance, the majority have neglected the utilization of error information, integrated the forecasting value of every component with simple ensemble approaches, and ignored forecasting stability, which may make the forecasting results poor. Considering the above drawbacks, a two-stage forecasting system is performed in our study based on the data preprocessing approach, improved multi-objective optimization algorithm, error correction and nonlinear ensemble strategy. The developed system effectively overcomes the shortcomings of previous models and observably prompts wind speed forecasting capacity. For investigating the prediction capacity of the developed system, six wind speed series obtained from China and Spain are applied as case sites. The forecasting consequences indicate that the developed system is more conducive to enhancing forecasting precision and stability than other involved models, which can provide beneficial assistance for wind speed forecasting. INDEX TERMS Wind speed forecasting, multi-objective optimization algorithm, ensemble strategy, forecasting accuracy.

Water quality forecasting has great practical significance for sustainable utilization of water resources and timely pollution prevention and control. However, owing to irregularity and volatility of water quality data, achieving accurate forecasts remains a challenging problem. Existing single forecasting

No abstract

Reliable photovoltaic and wind power generation forecasts are essential for efficient power systems operations. A combined forecasting system is developed, which integrates a data preprocessing method, a sub-predictor selection rule, and a multi-objective optimization to integrate various forecasting models. The proposed system effectively aggregates the advantages of all algorithms involved, facilitating greater prediction precision and stability. Experiments indicated that the proposed system can achieve higher quality point and interval forecasting performance relative to the comparative approaches.

No abstract