Research And Implementation Of Short-term Wind Power Prediction Based On Machine Learning

In recent years,the construction of wind power in China has been maintaining rapid growth.However,the intermittency and uncertainty of wind power bring great challenges to wind power grid connection.At the same time,the lag of grid construction leads to the serious phenomenon of wind abandonment.Accurate prediction of wind power can improve the stability of grid-connection and reduce the wind abandonment rate.After investigating and understanding the relevant algorithms of wind power short-term power forecast in recent years,this thesis starts from the reality of wind power forecast of the eastern Inner Mongolia and uses machine learning method SVM,bayesian algorithm and LSTM long and short term memory algorithm to solve the problem of wind power short-term power forecast of the eastern Inner Mongolia.In this thesis,evaluation criteria for regression models MAE and RMSE are used to compare the forecast results.According to the forecast results of three model experiments,LSTM algorithm is superior in short-term power forecast of the eastern Inner Mongolia.In order to make the algorithm research results of this subject have more practical value,the wind power output power forecast system is developed.The system uses Django framework,combined with Echarts visualization technology,to meet the visualization needs of users,wind power data for a variety of data interactive display.Through the development and construction of the short-term wind power output forecast system,not only the short-term wind power forecast of the eastern Inner Mongolia,but the visual interaction between wind power data and its users is realized.The test results show that the system is safe,stable,user-friendly and easy to operate,and can realize the prediction function in many ways.