Point and interval forecasting for wind speed based on linear component extraction

As a renewable energy, wind power attracts more and more attention. However, the intermittence and randomness of wind speed make the utilization of wind power a challenging task. Therefore, it is essential to improve the ability of wind speed forecasting. This paper proposes a new hybrid system for wind speed forecasting, which includes three modules: data pre-processing module, deterministic point prediction module and probability interval prediction module. Empirical mode decomposition (EMD) and singular spectrum analysis (SSA) are conducted to extract the linear component of the initial wind speed series in data pre-processing module, autoregressive integrated moving average model (ARIMA) and back propagation neural network (BPNN) are employed to produce the prediction points (PPs) of the initial data in deterministic point prediction module, and ARIMA and improved First Order Markov Chain (IFOMC) model are provided to make the uncertainty analysis of wind speed in probability interval prediction module. A case study is selected to test the performance of the new system. The simulation results demonstrate that the new system can achieve better precision and higher efficiency than several benchmark methods.