Short-term load forecasting and seasonal analysis of electrical energy consumption

Load forecasting, which is a critical step in efficiently planning, economically producing, distributing, and managing system operations of electric energy systems, as well as predicting future energy needs, holds significant importance in the electricity sector. These forecasts rely on predicting future electricity demand by examining consumption patterns from past periods. Load forecasting can be categorized into three different periods: short-term, medium-term, and long-term. Short-term load forecasts are made for a short period such as a day or a week; medium-term load forecasts cover a longer period, such as several months or a year, while long-term load forecasts extend to several years or more. In this study, short-term load forecasting modeling was conducted using hourly electricity consumption data from a facility located in the Southeast Anatolia Region for the years 2019-2020-2021-2022. Additionally, temperature values for these years were obtained from meteorological data, and a temperature factor was added to the electricity consumption data. Seasonality analysis was performed to evaluate the impact of the temperature factor on the results of electricity consumption forecasts. For load forecasting application in this study, regression, artificial neural networks, and ARIMA methods were utilized. Forecast results were obtained for each method and compared with other methods. In the comparison made in terms of data accuracy and analysis speed, average error values were determined, and it was observed that the ARIMA method provides more accurate predictions by reducing fluctuations in load data. The artificial neural networks method, on the other hand, was found to produce results more quickly by learning patterns in load data.