Design of a combined wind speed forecasting system based on decomposition-ensemble and multi-objective optimization approach

Wind-speed forecasting plays an important role in the efficient utilization of wind energy. However, accurate and stable forecasting of wind-speed series is challenging, considering the nonlinearity and chaotic characteristics of wind. Moreover, the limitations of individual forecasting models are ignored, which invariably leads to poor forecasting precision. Therefore, here, a wind-speed forecasting system based on two types of machine learning approaches (decomposition-ensemble and multi-objective optimization) is proposed, which addresses the nonlinearity and chaotic characteristics of wind-speed series well. In this system, the advanced optimization algorithm and no negative constraint theory determine the weights of results decomposed and forecasted by the sub-models. An empirical study using 10 min and 30 min interval datasets shows that the combined forecasting system outperforms comparison models and has advantages for wind-speed forecasting. (C) 2020 Elsevier Inc. All rights reserved.

Automated summary

Wind-speed forecasting is crucial for the efficient utilization of wind energy, but accurate prediction is challenging due to nonlinearity and chaotic characteristics. A machine learning-based forecasting system incorporating advanced optimization algorithms and no negative constraint theory has been proposed, showing superior performance in empirical studies.