The study and application of a novel hybrid forecasting model - A case study of wind speed forecasting in China

Given the current increasingly serious energy crisis, the development and utilization of new energy resources are attracting increasing attention, and wind power is widely used among these renewable energy resources. However, the randomness of wind power can cause a series of problems in the power system. Furthermore, the integration of large-scale wind farms into the whole power grid can place a great burden on stability and security. Accurate wind speed forecasting would reduce the randomness of wind power, which could effectively alleviate the adverse effects on the power system. In this paper, a hybrid wind speed forecasting model is proposed with the hope of achieving better forecasting performance. Wavelet Packet Transform (WPT) was employed to decompose the wind speed series into several series with different frequencies. A Least Square Support Vector Machine (LSSVM), the parameters of which were tuned by a particle swarm optimization based on simulated annealing (PSOSA), was built to model those series. The optimal input form of the model was determined by Phase Space Reconstruction (PSR). To verify the effectiveness of the proposed model, the daily average wind speed series from four wind farms in Gansu Province, Northwest China, were used as a case study. The results of the simulation and Grey Relational Analysis indicate that the proposed model outperforms the comparison models, and the null hypothesis of the predicted series having the same mean of the real series was accepted.