Regression models in wind power forecasting

Modeling of generation of wind power systems is useful for an effective management and balancing of a power grid, supporting real-time operations. Forecasting the expected wind power production could help to deal with uncertainties. In comparison with the mathematical approach, the data driven approach is useful where both detailed information about the system and real time measurements are unavailable. Winds being a natural phenomenon, statistical methods are more suitable for wind power plants than that of conventional power plants. In this paper, the data on the wind speed and power generated from a location in the state of Karnataka, India, has been analyzed and shown that the probability distribution of wind speed follows Rayleigh or Gaussian/Normal distribution. Short-term wind power forecasting is carried out using Autoregressive models.