Short-Term Electricity Load Forecasting and Seasonality Analysis Using Temperature and Artificial Intelligence Methods in the Southeastern Anatolia Region

Abstract

The planning of electrical energy systems can be realized in a more efficient and sustainable way by forecasting energy demand accurately. In this context, short-term load forecasting plays a critical role in optimizing energy production and distribution processes. In this study, short-term load forecasting was conducted using hourly electricity consumption data from a facility located in the Southeastern Anatolia Region between 2019–2022. The data were integrated with meteorological parameters to evaluate the impact of temperature. The performance of Multiple Linear Regression (MLR), Artificial Neural Networks (ANN), and AutoRegressive Integrated Moving Average (ARIMA) methods were compared. According to the results, the ARIMA method was the most successful with an accuracy rate of 92%, followed by the ANN model with 90% accuracy. The MLR method demonstrated relatively lower performance, achieving an accuracy rate of 89%. Moreover, ANN showed a strong capability to model complex relationships, while ARIMA excelled in datasets with seasonality. In conclusion, this study highlights the strengths and weaknesses of different methods, providing valuable contributions to energy planning and emphasizing the importance of analyses conducted using regional datasets.