Short-Term Wind Speed Forecasting Based on Singular Spectrum Analysis, Fuzzy C-Means Clustering and Improved SSABP

Overuse of non-renewable energy has seriously affected environment, wind power is of vital significance to alleviate the energy crisis and protect the environment. The forecasting accuracy of wind speed has a positive correlation with the effective utilization rate of wind energy. BP neural network has advantages over traditional prediction models in dealing with nonlinear problems, but it has different generalization ability to different data. Therefore, we proposed a hybrid forecasting model based on singular spectrum analysis (SSA), fuzzy c-means clustering (FCM) and improved sparrow search algorithm (ISSA-BP). Firstly, after the raw wind speed data are obtained, singular spectrum analysis is employed to de-noise, so as to improve the data quality. Secondly, the input dataset of BP is divided into several categories using FCM, and the number of classifications for two different datasets is obtained through multiple experiments. Thirdly, since the selection of parameters largely determines the performance of BP neural network, the improved sparrow search algorithm (ISSA) is adopted to optimize the weights and thresholds. Then different ISSA-BP models are built for each class of input datasets. Finally, the class of forecasting input is determined and the corresponding ISSA-BP is used for forecasting. The experimental results of two cases showed that the proposed model was not only suitable for one-step forecasting, but also improved the accuracy of multi-step forecasting. The ultimate experimental results, in comparison with other six different models, demonstrate that the proposed model can acquire higher prediction precision.