Forecasting of Wind Speed in Sivas Province and Surroundings with Long Short Term Memory Machine Learning Method

Sungur H., Çıtakoğlu H., Aktürk G.

5 th International Artemis Congress on Life, Engineering, and Applied Sciences, İzmir, Turkey, 1 - 03 October 2023, vol.1, pp.1-2

  • Publication Type: Conference Paper / Summary Text
  • Volume: 1
  • City: İzmir
  • Country: Turkey
  • Page Numbers: pp.1-2
  • Erciyes University Affiliated: Yes


Wind is one of the important effects used in the field of meteorology. It is very important to know wind speed because wind uses in clean energy generation, agricultural calculations, calculations of load acting on structures and having a low carbon footprint, being sustainable and provides energy independence. General Directorate of Renewable Energy foreseen will be increased the share of renewable energy resources in total electricity generation to 30% in 2023 in the 2013-2023 Period Turkey National Renewable Energy Action Plan. The share of renewable energy generation in total energy generation in 2021 is 19.1% according to the Turkish Statistical Institute. The rate of wind power in licensed total electrical energy generation was stated as 9.71% in the 2021 report of the Energy Market Regulatory Authority. Accurate wind speed forecasting is required in order for renewable energy sources to reach the projected share. According to the Turkish Statistical Institute, based on Turkey’s economic gross domestic product, the ratio of the agriculture and forest sector to the total of sectors is approximately 6% for 2021. This rate is approximately 16% for Sivas province. The fact that Sivas province’s share of gross domestic product in the agriculture and forestry sector is above Turkey’s average shows the importance of evapotranspiration and the related wind speed value in Sivas province. For this reason, in the study, a wind speed forecasting study was carried out with the Long Short Term Memory machine learning method, using the wind speed values belong the stations in Sivas province and its surroundings. Wind speed values of Sivas Center, Suşehri, Kangal, Gemerek and Divriği stations located in Sivas province were obtained from the General Directorate of Meteorology. The data set consists of 432 data between the years 1986-2021. 75% of the data was used in training the model and 25% was used in the testing phase. Mean absolute error (MAE), root mean square error (RMSE) and coefficient of determination (R2 ) performance criteria were used to determine the model performance. MAE values were calculated for each station, respectively 0.1348; 0.6848; 0.6075; 0.2266; 0.2803; RMSE values were calculated 0.1400; 0.7539; 0.6377; 0.2618; 0.3238; R2 values were calculated 0.9917; 0.9837; 0.9870; 0.9878; 0.9957. The Long Short Term Memory model performed a good predictive model for wind speed in generally. Keywords: Wind Speed, Long Short Term Memory, Turkey, Forecastin