Forecasting Model of Electricity Production from Hydroelectric Sources with Long Short-Term Memory (LSTM) Networks

Abstract Electricity is one of the most important elements for economic growth and development of societies in today’s modern societies. The research of electricity generation, knowing the size of the electricity supply, and the methods developed to meet this supply are among the important subjects of study today. With the increase in electricity supply and the increasing importance of environmental pollution, the use of renewable energy sources in electricity generation is increasing. In this study, Long Short-Term Memory (LSTM), a type of recurrent neural network, is used to predict the energy production in a hydroelectric power plant. The LSTM method is one of the most popular recurrent neural network methods and is widely used in the field of deep learning. The graphical and numerical results obtained at the end of the study show the success and efficiency of the LSTM method. Ct represents the updated cell state. With ft , forgotten information is removed, with i t , new information is added. In the last step, the output layer is obtained by using the equations given below. Cite this article as: İ. Özge Aksu and T. Demirdelen, “Forecasting model of electricity production from hydroelectric sources with long short-term memory (LSTM) networks,” Turk J Electr Power Energy Syst., 2024; 4(3), 159-164.