Abstract:

Randomness, intermittence and fluctuation are features of new energy, which includes wind energy and solar energy, and power forecasting is an effective solution. The characteristics of wind power output and forecasting model are fully considered to propose piecewise support vector machine (PSVM) and neural network (NN) model; the effort of weather condition on photovoltaic is analyzed to optimize the forecasting model. The case studies from several wind farms and photovoltaic power stations prove that the proposed models have higher precision, which offer support for reliability analysis of power output forecasting.

Key words: wind power, photovoltaic, power forecasting, support vector machine, neural network, wavelet analysis