Short Term Wind Speed Forecasting And Confidence Imterval Test Of Wind Speed Predicition

With coal, oil and other conventional energy sources are exhausted, the world insparing no effort to find alternative energy sources, develope and use the new energysources. At present, the main way of exploitation and utilization of wind energy is thelarge-scale wind generation connected with grid. More than 30% annual growth ininstalled capacity. However, the difficulty of wind power dispatch is increased, due tothe randomness and uncontrollable of the wind power and uncontrollable.timely andaccurate wind forecasts can reduce operating costs and power system spinning reserve,then can reduce the impact on the grid and improve the efficiency of wind power.The size of wind speed forecast error is connect with the wind speed predictionmethod and the characteristics of wind speed distribution. Fristly, Analysis thefunction of the wind speed distribution, the way for finding wind speed distributionfunction parameters, the rule of wind speed changes, the relation between wind andwind power,and then study the time series characteristic forecasting model. Theninstroduce wind speed time series characteristics and forecast model .secondly,introducing weights facts on the basic of the least squares support vectormachine .using adaptive parameter optimization method to adjust the parameters ofLS-SVM, giving a different weighted acoring to different sample, adaptive adjustmentof weighting coefficients to reduce the error, then established a good predictionmodel.To effectively use the results of wind speed and wind power forecasting, on thebasis of the prediction model which is combined by wavelet decomposition and leastsquares support vector machine, The confidence level of the prediction value has beenverified according to the distribution characteristics of wind speed and the confidencetheory so as to avoid the decision-making risk caused by single larger prediction error.