The Research On Wind Speed And Wind Power Forecasting Of Wind Farm

The research on wind power is deeply and widely at home and abroad. But windspeed forecasting and wind power forecasting in wind farm still can’t make ussatisfied. This paper makes research and discussion in the field. When Wind PowerPenetration is over a given ratio, it will severely influence power quality and theoperation of power system. If wind speed or wind power in wind farm can beforecasted accurately, when necessary, dispatching department can adjustdispatching plan according to the forecasted data ahead of time. So it will lessen theunfavorable influence caused by wind power to the entire electrical network.Moreover it has other significance to power system.On the basis of others’ research, puts forward three methods based on SVM(support vector machine) forecasting of wind speed, and got less than 25% of theMAPE. Firstly, choose LSSVM (Least Squares Support Vector Machine) which iswildly applied in function estimation and approximation. Then introduced GA(Genetic Algorithm) to optimize the parameters choice of the SVM, introduced EMD(Empirical Mode Decomposition) divided the non-stationary wind time series intoseveral stationary wind time series. This article describes three optimization methodbased on support vector machine model for wind speed forecasting. There areLSSVM (Least Squares Support Vector Machine), GA-LSSVM (Genetic Algorithmsbased Least Squares Support Vector Machine), and EMD-GA-LSSVM (Empiricalmode decomposition combined with GA-LSSVM). Compare the effect of wind speedforecast for three methods with three different forecast periods. Finally, forecastresult of wind speed was converted into wind power prediction results based on windpower curve.From what has been discussed above，the forecast models were discussed indepth by using support vector machine, data processing and numerical calculationwas carried out, and the forecast results of different combination models have theirown characteristics.