Wind Power Prediction Based on Difference Method

Wind Power Prediction Based on Difference Method

The renewable wind power sources are difficult to be predicted in view of the fluctuating factors such as wind bearing, pressure, wind speed, and humidity of the surrounding atmosphere. An attempt is made in this paper to propose a difference method to build a neural network and a long short term memory (LSTM) model for wind power prediction. First, the correlation of each data is analyzed and then per-forming difference processing on the original data to solve the problem that the original data cannot be analyzed by probability distribution. The prediction is made by building the neural network and LSTM and feeding the original data and the difference-processed data into the neural network model respective-ly. Finally, the data are added for validation, and the raw data used include wind power data in Belgium from November 1, 2019 to November 30, 2019.The experimental results show that the LSTM prediction accuracy is improved by 178.67%, and is effective in predicting long-term wind power data with 216.06% accuracy improvement, the neural network prediction accuracy is improved by 154.07%, and the short-term wind power prediction accuracy is improved by 228%.