Short-term Forecast Of Output Power On The Wind Farm

The main way of exploitation and utilization of wind energy is wind generation andinjecting large-scale wind power into power grid. Wind power is an intermittent energy,which has strong randomness and instability. Because wind power varies with cube ofwind speed when the speed is between cut in and rated value, the fluctuation range ofoutput power is very large. These problems present challenges for security operationand reasonable scheduling of power grid. Wind power prediction is an effectivemethod to solve these problems. It is in favor of electricity market trading, transientstability assessment, and as a basis for the stable operation and scheduling to reducethe harmful effects to power grid.Based on different data used by forecasting models, wind power forecast can bedivided into two categories. The first is the direct forecasting methods which useoutput power time series. The second is the indirect forecasting methods which usenumerical weather prediction data, e.g., wind speed. According to time scales, windpower forecast include short-term forecast, medium-term forecast and long-termforecast. The main study content of this paper is wind power short-term forecast.Based on two single prediction models, an optimal variance combined model isproposed. Two single models are built based on weather data and output power timeseries, respectively. So the combined forecasting model is able to use all effectiveinformation of different data. Using genetic algorithm to optimize the parameters ofartificial neural network, and then based on optimized network to build combinedforecasting model. V-system is used to deal with non-stationary time series. A windpower forecasting model based on V-system and artificial neural network is proposed.A wind speed forecasting model based on V-system and least square support vectormachine is built. According to wind speed forecasting values, considering theinfluencing factors, output power forecasting model can be established.