Research On Short-term Wind Power Forecasting Based On DA-LSTM

With the increasingly prominent energy crisis and environmental problems,renewable energy has become an increasingly important part of the global energy structure.As a clean renewable energy,wind energy is currently the new energy with the greatest development potential and the widest application prospects.However,the volatility and intermittency of wind energy hinders its further development in the electricity market,and at the same time poses a challenge to the resilience and reliability of the power grid of China.The power system can use accurate and effective wind power short-term prediction technology to optimize wind power dispatch and operation management while absorbing more wind power.In order to improve the availability of wind energy and the accuracy of the prediction model,this paper explores the application effect of the DA-LSTM model in the wind power short-term prediction based on a comprehensive analysis of historical wind power data.The main research contents are as follows:(1)In order to ensure the reliability and completeness of the historical data set,the data is preprocessed.First,the method of combining isolation forest and linear interpolation is used to detect the outliers and fill the missing values of the original data,and then use the discrete standardization method to normalize the data.The results show that the method can effectively deal with abnormal data in historical data,and the repair of abnormal data and the filling of missing data significantly improve the usability of data.In addition,the complete data set is normalized to improve the learning efficiency of the predictive model.(2)In order to improve the learning efficiency of the training model,the correlation analysis between the climate characteristic variables and wind power in the pre-processed data is carried out.Through comparative analysis of the Pearson coefficient values between different climate characteristic variables and wind power variables,the wind speed variable with the strongest correlation with wind power is retained,and the data of other variables are screened out.(3)In order to improve the performance of the wind power short-term prediction model,a wind power short-term combined prediction method based on DA-LSTM is proposed.This method first uses the dragonfly algorithm to optimize the internal parameters of the LSTM neural network.The LSTM neural network can use the optimal parameter values to achieve short-term prediction of wind power.In addition,in order to verify the superior performance of the proposed prediction model,three sets of comparative experiments are designed in this paper.Based on the three sets of experimental results,it can be seen that the dragonfly algorithm is effective and feasible for optimizing the parameters of LSTM neural network,and the DA-LSTM prediction model can be well adapted to prediction scenarios with large data scale and high data complexity,and can also be applied In different seasons of wind power prediction scenarios,and in the above scenarios,DA-LSTM can maintain high prediction accuracy and good prediction performance.