Research On Prediction Of Wind Power Based On Non-parametric Regression Algorithm

Nowadays,as people's demand for energy continues to increase,new energy sources are vigorously developed with advantages of cleanness and renewable.Among them,the development and utilization of wind energy is favored by various countries in the world.At the moment,the total installed capacity of wind power in China has been among the best in the world,but there is a problem of inefficient use of wind resources.Therefore,how to improve the efficiency of wind resource utilization is still an urgent problem to be solved.An accurate prediction of future wind power can make the subsequent scheduling more scientific and reasonable,so as to improve the efficiency of resources utilization.Based on this background,this paper has carried out a series of researches to improve the prediction accuracy of wind power.In this paper,firstly,based on the original data of wind power site,the error iterative algorithm is used to process the original data,and the abnormal data samples are eliminated.Then,the estimation method of the total quantile in the non-parametric estimation method is used to denoising,in which the data with denoising with 0.9 as the confidence interval is the best.Secondly,based on the nonparametric regression algorithm,after selecting Gaussian kernel function fitting function and fixed window width,the paper normalizes the data and adopts three error evaluation systems: R-square,Root Mean Square Error and Mean Absolute Percent Error.And then,the interpolation method is used to process the missing data,and the data is input into the nonparametric regression model,which is concluded that the wind power and the actual wind power curve are well fitted,and a 4-hour power forecast is carried out,which has a better prediction effect.In order to further improve the prediction effect,this paper,based on fixed window width,analyzes and contrasts the four kernel functions of Gaussian kernel function fitting,Epanechikov kernel function,Quadratic kernel function and Sigmoid kernel function,and the results show that the quadratic kernel function is the best one.At the same time,based on the Quadratic kernel function,the ant colony optimization is used to optimize the window width,and find the window width value suitable for the wind turbine research in this wind power site.After selecting the Quadratic kernel function and the optimal window width,the data is input into the nonparametric regression model,the fitting and prediction accuracy are further improved.The research results show that the proposed nonparametric regression algorithm can predict wind power and has a good effect,which provides a reference for improving the utilization efficiency of wind resources.