Research On Short Term Wind Power Prediction Of Large Scale Wind Farm

Wind energy has randomness and instability,which is not conducive to large-scale grid-connected wind power,reduce the advantage of wind power bidding.In order to improve the stability of wind power grid,improve the scale of wind power into the grid,reduce the cost of wind power operation,we must improve the prediction precision of wind speed and wind power.In this paper,the indirect prediction method is used to predict the wind speed by using the neural network combination method based on the general average empirical mode decomposition method,and the power prediction is carried out on the basis of the corresponding wind speed-power curve.Firstly,the variation of wind speed has seasonal characteristics,and the variation characteristics and distribution characteristics of wind energy are analyzed in different seasons.The variation characteristics of wind speed include diurnal and monthly variation characteristics,the characteristics of diurnal variation in different seasons are presented,the Weibull curve of wind speed is established,and the shape parameters and scale parameters affecting the distribution in different seasons are analyzed.In addition,the distribution characteristics of wind direction and wind speed,direction and power are analyzed.To establish wind speed as the main research object of power prediction.Secondly,according to the instability of wind speed series,a neural network combined wind speed forecasting method based on empirical mode decomposition method is proposed.This paper analyzes the endpoint effect and modal aliasing of empirical mode decomposition method,uses the image continuation method to advance the data sequence,and uses Ensemble Empirical Mode Decomposition method to decompose the wind speed sequence.The decomposed components are modeled and predicted by least squares support vector machines,and the final prediction results are compared.According to the different fitting characteristics of the sequence of wind speed decomposition,the complexity and randomness of the intrinsic component 1 are separately predicted by LSSVM,RBF and GRNN respectively,and the GRNN network is used to predict the predicted results,and the other components are modeled as the other components using the LSSVM modeling prediction,The predicted results are added and obtained,and the predicted results are compared.Finally,under the different temperature and humidity conditions,different wind turbine units,wind power curve will produce a certain difference,so it sets up different wind power curve on different seasons and different wind turbine wind power curve,and in the existing forecast wind speed based on the prediction of air power.Wind power forecasting accuracy to meet the requirements of wind power network.