Research Of Wind Power Generation Ultra-short-term Wind Speed Prediction Model Based On Support Vector Machine

Energy issues have become a global problem. The wind as a clean and renewable energy, its capacity and power generation are constantly improving, but the wind has volatility, instability characteristics and systems will bring bad impact for network power. Power system’s security, stability and efficiency have been greatly affected.In order to reduce the adverse impact for wind power and reasonable scheduling of wind resources, accurate predict the wind energy resource is very necessary in the farm.For the long-term wind speed prediction, there are already many scholars involved in this area,while the predict result is very well;But for the short and ultra short-term wind forecasting research has not yet reached very good results, mainly due to the wind randomness and non-stationary characteristics caused.First,this article do a brief introduction for the wind power industry development status and background, the existing wind speed prediction model approach is described in detail.Second, the main research is based on support vector regression model to predict the wind speed, introduces the support vector machine theory and support vector regression principles and their implementation issues.According to online data to predict one-period wind speed, and then extended to predict multiple periods, extending the wind forecast time and success predict a certain point wind speed.Finally, the principal component analysis theory and particle swarm optimization, and support vector machines combine organic. Achieving a certain surface wind speed prediction, improved the wind forecasting accuracy.At last,the paper concludes a number of shortcomings and explores some new methods.