Short term wind power forecasting using Hilbert-Huang Transform and artificial neural network

Capacity and output power forecasting have great significance in seamless integration of renewable energy to the grid. However, the uncertainty of wind power and intermittence of wind energy are the main factors which affect forecasting precision. The wind power output data can be treated as a signal stream which has characteristics for possible wind capacity forecasting. Hilbert-Huang Transforms (HHT) and Hilbert spectral analysis have been applied extensively to analysis nonlinear and non-stationary stochastic signal. The time series of wind power output has been transformed into certain signals with different frequencies. Each signal is taken as input data joining with wind speed data to establish Artificial Neural Network (ANN) forecasting model. The models are combined together to obtain the final results on potential wind power output. This paper proposes HHT-ANN model for wind power forecasting. A case study of a wind farm in Texas, U.S shows that the MRE of proposed method is lower than the traditional ANN approach.