Research On The Short-Term Prediction Of Wind Power Output

Rapid global economic and social development, energy supply and demand andenvironmental pollution problems have become increasingly prominent, developmentand use of clean resources to generate sustainable development of individual countriesbecome consensus and energy strategy. Since a huge amount of wind energy resourcesand clean, renewable, so the use of wind power is a kind of broad prospects forsustainable energy use. Wind power is different from the conventional power ofrandomness and other characteristics, wind power capacity increased significantly affectpower system security, stability and economic operation. Achieve higher accuracy in favorof wind power output forecast grid scheduling and network management, and to ensurethe safe operation of the system, improve the system capacity of wind power and windpower consumptive bidding capabilities.Firstly, the development of wind power industry at home and abroad as well as windpower output prediction technology research status summary and conclusion. Taking intoaccount the statistical method is wind power output and short-term forecasting of themost widely used prediction methods, and time series forecasting model and supportvector machine model is a statistical method more typical prediction model, this paperbased on time series method and support vector machine theory respectively.established based on the time series method based on support vector machine and windpower output short-term forecasting model, using the built forecasting models areexamples and comparative analysis of the characteristics of two prediction models.Traditional statistical forecasting methods by building between input and outputparameters of the function to make predictions work due to the randomness of windpower output and complexity, the establishment of parametric predictive model is notnecessarily the best choice. Through further analysis of wind power output prediction ofstatistical methods in prediction idea and combining nonparametric regression principle,the use of pattern recognition clustering algorithm, this paper presents a wind powerprediction mode based on pattern recognition and nonparametric regression (PRNPR)and the performance of the model was tested. with the performance of the twoprediction models were compared, the results show that the proposed prediction modelbased PRNPR has better prediction accuracy, and easy modeling, there are certain practical value.Different prediction models as well as the theoretical basis of information modelingusing different, but the linkages between each other, complement each other. Finally, thispaper based on the idea of information fusion of different prediction methods by fusingthe information contained in the improved PRNPR established model for predictingexample shows that using the maximum entropy principle to improve predictive modelsto increase the robustness of the predictive model, to further improve the prediction ofwind power output accuracy.