Energy Production Prediction in Hydroelectric Power Plants with Multi-Layer Perceptron Algorithm, Menzelet Dam Example

• Published in: International Journal of Applied Methods in Electronics and Computers
• Main Authors: İSPİR, Ezgi Öztürk, KOÇER, Hasan Erdinç, KUMCU, Şerife Yurdagül
• Format: Journal Article
• Language: English
• Published: 30.06.2025
• ISSN: 3023-4409; 3023-4409
• DOI: 10.58190/ijamec.2025.122

Hydroelectric energy is connected to clean and extractable energy produced by electric generators that convert the movement of water falling into dams into energy. From the perspective of life cycle planning of energy production systems, estimating the energy to be produced from hydroelectric power plants is very important in terms of energy production efficiency management, but it is quite difficult to do. Because the flow of such energy production data depends on factors such as precipitation-flow, flow, temperature and evaporation. This causes energy changes and fluctuations in variables. In this paper, long-term energy production planning was made using Multi-Layer Perceptron (MLP) from Artificial Neural Network architectures for Menzelet Dam and HEPP located in Ceyhan Basin of Kahramanmaraş province. The activation functions used in this study are sigmoid and tanh and the models used for learning are quick propagation and conjugate gradient descent. In the study, the energy production data between (1999-2020) is used for the experiment. The training and test parts were run. The results of the prediction values were compared by looking at CCR and R2 values. According to the tests, the highest prediction value for energy is 0.9891. Hydroelectric energy is connected to clean and extractable energy produced by electric generators that convert the movement of water falling into dams into energy. From the perspective of life cycle planning of energy production systems, estimating the energy to be produced from hydroelectric power plants is very important in terms of energy production efficiency management, but it is quite difficult to do. Because the flow of such energy production data depends on factors such as precipitation-flow, flow, temperature and evaporation. This causes energy changes and fluctuations in variables. In this paper, long-term energy production planning was made using Multi-Layer Perceptron (MLP) from Artificial Neural Network architectures for Menzelet Dam and HEPP located in Ceyhan Basin of Kahramanmaraş province. The activation functions used in this study are sigmoid and tanh and the models used for learning are quick propagation and conjugate gradient descent. In the study, the energy production data between (1999-2020) is used for the experiment. The training and test parts were run. The results of the prediction values were compared by looking at CCR and R2 values. According to the tests, the highest prediction value for energy is 0.9891.