Short-Term Electricity Load Forecasting Using Artificial Intelligence (Artificial Neural Networks) Method and Meteorological Data for Sivas-Central 1st Organized Industrial Zone

The effective use of electricity, accurate prediction methods are becoming important for energy planning and prevention of imbalance costs. Efficient, high-quality, and safe electricity supply processes are also crucial for maintaining the supply-demand balance in the electricity market. Therefore, it is necessary for electricity producers and suppliers to make good planning, and there is a need to predict unrealized consumption values for planning activities. The aim of this study is to make short-term electricity load forecasting for Sivas-Central 1st Organized Industrial Zone using meteorological data and past consumption values. It has been observed in the literature that artificial intelligence, especially Long Short Term Memory (LSTM) architectures, are effectively used in the energy forecasting field. In this study, an LSTM architecture was created using Matlab/Simulink to make load forecasts. In this architecture, four cases are used using different data sets for input data. In the first case, only days of the year are used as input data sets. For the second case, “Month, Day of Month, Working Day and Days of Year” are used. For the third case, additional “Holiday Days, Holiday, Weekend, Normal Day, Dollar” data is used in the second case. In the fourth case, additional “Meteorology” data is used in the third case. As a result, the RMSE values for the four cases are as follows; RMSE1: 1501.897, RMSE2: 1005.312, RMSE3: 1343.179, RMSE4: 9342.352. This study aims to provide effective solutions in supply-demand balance and energy management. This study deeply analyzes the effect of different data combinations on LSTM-based consumption estimation and offers an original contribution in terms of showing that using more data does not always provide better estimation accuracy. The aim is to provide significant contribution to energy suppliers and industrial zones for accurate energy management through the predictions made.