(Short-Term Wind Speed Estimation with Deep Learning Method: Bingöl Example)

Energy consumption is increasing day by day all over the world due to developments in industry and technology. In addition to being exhaustible resources, fossil fuels have negative environmental effects. For this reason, countries are forced to turn to alternative resources to meet the increasing energy demand. Renewable energy sources, especially solar and wind, can be a good solution to meet energy needs because they are endless resources that do not pollute the environment. However, since they are constantly variable by nature, continuous energy production at a fixed value is impossible with these sources. Energy producers need to be able to accurately estimate the energy that can be produced from wind in advance. Predicting the production to be made with a very small margin of error is necessary for production planning. However, it is not easy to estimate the wind speed, which changes depending on many factors, especially humidity and temperature, and the amount of energy to be obtained accordingly. Recent studies have shown that deep learning methods can be a good solution for estimating wind power and wind speed. Wind speed data from previous periods of a region can be trained with deep learning architectures and future wind speed estimations can be made. In this study, the wind speed data obtained from the Bingöl Meteorology Directorate and measured between 01.01.2020 and 01.02.2021 for Bingöl province were processed with the LSTM (Long Short-Term Memory) model, which is one of the deep learning methods, and a one-hour wind speed estimate was made. The study provides information about the region’s wind potential and the wind power that can be produced for energy investors and researchers.