Forecasting wind power generation using machine learning techniques: a case study

Akalın H., Erdoğan Z., Dereli T., Altuntaş S., Eroğlu Y.

International Conference on Artificial Intelligence towards Industry 4.0, Hatay, Turkey, 14 - 16 December 2019, pp.1-2

  • Publication Type: Conference Paper / Summary Text
  • City: Hatay
  • Country: Turkey
  • Page Numbers: pp.1-2
  • Yıldız Technical University Affiliated: Yes


Energy is the essential input of production processes, economic development, and social welfare. One of the most critical factors for sustainable development is access to uninterrupted energy. There are several ways to supply energy demand, such as fossil fuels, renewable energy sources, nuclear plants, etc. Each country has different energy strategy plans due to her energy demand, which has to be supplied securely and clearly in view of climate change and energy security. While the energy needs of the countries are increasing day by day, traditional energy resources have difficulty in meeting this increasing demand. To cope with that problem, renewable energy sources should be forecasted for sustainable energy development and environmental protection. Wind energy is one of the most prominent renewable energy sources in the world. However, wind power generation varies from other traditional energy generation.  It is not easy to generate wind power for power system operators due to the unpredictability and nonstationary nature of wind power generation. Traditional forecasting methods can be insufficient with these uncertainties and risks.  The aim of this study is twofold.  Firstly, it attempts to estimate wind power generation. Secondly, it analyzes the effects of different input parameters on the power generation by using weekly averaged data from an operated wind farm in Turkey. This study uses two different machine learning techniques, namely the Multi-Layer Perceptron and Bagging Algorithm. The results of this study show that the ensemble machine learning methods give better results than individual methods. The forecasting model established in this study also allows determining the most critical factors affecting the amount of wind power generation, such as wind and temperature values. Decision-makers should attach importance to these factors to increase the amount of wind power generation.