Short-term Forecast of Wind Speed Based on Hybrid Multi-Stage Decomposition and LSTM Optimized by an Improved WOA

Wind power, which is a widely recognized clean and renewable resource, has attracted extensive concerns in the worldwide. It is also frequently regarded as the substitute of fossil fuels for its economic and pollution-free characteristics. However, the stochastic character of wind speed brings serious threat to the management and utilization of wind power system. Therefore, wind speed forecasting methods with high accuracy are significant for economically integrating wind power energy. In this paper, a novel forecasting technology is proposed based on secondary decomposition technology (SD) and long short-term memory (LSTM) trained by an improved whale optimization algorithm (IWOA), which considerably decreases the instability of the initial data and improves the convergence accuracy of the algorithm. The proposed method focuses on point forecast and the performance is demonstrated by a series of experiments based on a real-world wind speed dataset. Specifically, the superior accuracy and stability of the proposed method are achieved compared with other benchmark models.