A hybrid forecasting approach applied in wind speed forecasting based on a data processing strategy and an optimized artificial intelligence algorithm

Conducting the accurate forecasting of wind speed is a both challenging and difficult task. However, this task is of great significance for wind farm scheduling and safe integration into the grid. In this paper, the wind speed at 10 or 30 min is predicted using only historical wind speed data. The existing single models are not enough to overcome the instability and inherent complexity of wind speed. To enhance forecasting ability, a novel hybrid model based on complementary ensemble empirical mode decomposition (CEEMD) and modified wind driven optimization is introduced for wind speed forecasting in this paper. CEEMD is utilized to decompose the original wind speed series into several intrinsic mode functions (IMFs), and each IMF is forecasted by back propagation neural network (BP). A new optimization algorithm combined Broyden family and wind driven optimization is presented and applied to optimize the initial weights and thresholds of BP. Finally, all forecasted IMFs are integrated as final forecasts. The 10min and 30min wind speed from the province of Shandong, China, were used in this paper as the case study, and the results confirm that the proposed hybrid model can improve the forecasting accuracy and stability.