Abstract:

Modeling of wind power is essential for an effective management and balancing of a power grid, supporting real-time operations. Forecasting the expected wind power production would help to deal with uncertainties. The data driven approach for forecasting is expected to give detailed information on the system and real time measurements. Wind being a natural phenomenon, probabilistic methods need to be employed in generated wind power, based on previous history of the system. In this paper, the data on the wind speed and power generated from a location in the state of Karnataka, India, has been analyzed for the duration of three months. It has been shown that the probability distribution of wind speed conforms closely to Rayleigh distribution. It is expressly demonstrated that, while the wind speed conforms to Rayleigh distribution, the electrical power developed follows a Weibull distribution in two parameters. Besides using graphical methods for estimating the Weibull parameters, Maximum likelihood equations are set up to estimate the parameters. These parameters have been used in estimating / forecasting of wind power using both Weibull algorithm as well as the Monte Carlo Method.