| Title |
A Study on Prediction of Wind Power Based on Deep-Learning Using Weather Data |
| Authors |
김은지(Eun-Ji Kim) ; 이택기(Taeck-Kie Lee) ; 김규호(Kyu-Ho Kim) |
| DOI |
https://doi.org/10.5370/KIEE.2021.70.5.735 |
| Keywords |
Deep-Learning; LSTM; Prediction; Wind Power |
| Abstract |
Although the proportion of wind energy sources in the system is increasing, the variability and stochastic characteristics of wind power have a negative effect on system stability and on the rescheduling of the system’s power generation. Therefore, an accurate prediction of wind power generation is necessary for stable power distribution in the power system. High-precision wind power forecasting provides a reliable basis for dispatching of power system. Since wind power data is time series data with uncertainty, a suitable model, the LSTM algorithm of Keras, was proposed to predict a short-term wind power generation. The LSTM has memory to store past states, so it is suitable for the problem of predicting time series data. This paper proposes a forecasting procedure using an LSTM neural network to forecast wind power. First, Pearson correlation coefficient method is utilized to determine the parameters of LSTM forecasting model. Then a case study was performed using real data collected from a wind farm in Yeongheung to confirm that the LSTM model was suitable for wind power prediction |