Research On Wind Speed Prediction Based On Support Vector Machine And Empirical Mode Decomposition

The desertification of land marked by wind sand activities and the disasters caused by dust storms are important environmental issues in the 21 st century.While solving the dynamics of this typical gas-solid two-phase fluid,it faces common problems such as the nonlinearity of the Navier-Stokes equation,the randomness of turbulent motion and multi-scale correlation.Although there are many wind speed prediction models,there are still some areas that can be developed and improved,such as how to expand from a net wind field to a wind sand flow field.Based on the good generalization of support vector machine and empirical mode decomposition,this paper establishes a mathematical model for short-term wind speed prediction.It is expected to deepen the understanding of wind and sand movement and provide some data support for the study of quantitative wind and sand physics.The main research contents include:Firstly,it summarizes the general research on the prediction of wind movement and wind speed,and introduces the basic characteristics and statistical description of the turbulence in the near-surface atmosphere.Secondly,pre-processed the original data of wind speed in the sand environment.Based on the support vector machine regression,a quantitative short-term wind speed prediction model is proposed.The calculation results are basically consistent with the observation data,and the relative error is within 9%.At the same time,it is found that proper preprocessing of the original data can reduce the prediction error and improve the operation efficiency.Finally,the wind speed time series is processed and interpreted according to the empirical mode decomposition to obtain components of different time scales,and an improved support vector machine wind speed prediction model is proposed.The results show that the improved prediction model has significantly reduced relativeerror,higher accuracy and better generalization.The short-term wind speed prediction model based on support vector machine regression and empirical mode decomposition constructed in this paper predicts that the wind speed is basically consistent with the actual wind speed.The improved model effectively reduces the prediction error and is worthy of further research and application.