Research On Short-term Wind Speed Forecast Based On Artificial Intelligence Algorithm

From the perspective of energy development trends,wind is a clean and inexhaustible energy source.So wind power,as a green renewable new energy power generation technology,has been widely used in various countries.However,there are still some problems in the process of wind power generation.Disturbances caused by wind speed changes will have a serious impact on the safe operation of the power grid.Therefore,accurate forecasting of wind speed becomes particularly important.It can not only improve the stability of the grid,but also reduce economic losses,rationally arrange the operation of the grid,and promote the transformation and development of global energy.This paper will study the point prediction and interval prediction of wind speed,based on wind speed data information,and combine decomposition algorithms,variable selection methods,artificial intelligence models and statistical knowledge,and improve the traditional prediction methods through optimization algorithms to improve the accuracy of wind speed prediction.The main contents of this research are as follows:(1)A new VMD improvement method was proposed.Decomposing wind speed can fully explore its potential information characteristics to facilitate the prediction work.However,when using VMD to decompose,the number of decompositions is difficult to determine.If the selected number is not appropriate,it will cause poor decomposition results.So this paper constructed the minimum sample entropy criterion,which can be used to determine the number of VMD decompositions.(2)Constructed a combination forecasting model with higher accuracy.This paper selected the appropriate neural network model based on the analysis of wind speed data.Then the research used FOA algorithm and PSO algorithm to optimize the parameters of models to improve the prediction accuracy,and built the Elman-RBF combined model.(3)The new EPSO optimization algorithm was built.Aiming at the problem that the PSO algorithm easily falls into the local optimum,this paper used the continuous search ability of the EO algorithm to improve it,so that the algorithm can converge quickly while avoiding premature convergence as much as possible.Finally,the optimization effect of the algorithm is improved.(4)The Fourier distribution was used for interval prediction.Based on the deterministic prediction of wind speed,the training error distribution was analyzed,and the Fourier function was used for fitting,and then its quantile and confidence interval were calculated to obtain the interval prediction result of wind speed.(5)The FCBF algorithm was applied to multivariate screening.In order to predict more accurately,other environmental factors affecting wind speed should be considered,and the FCBF algorithm was used to screen multiple factors to obtain appropriate input varibles.