Short-term electrical load forecasting using artificial neural networks

Planning, operating, and controlling the electric power system presents both technical and financial challenges for the electric power company. A proper assessment of the current and future electric loads is also necessary for the development and operation of an electric power system, where electric load forecasting is utilized to estimate the energy required to meet the demand. Many methods were used to anticipate short-term electrical loads, but they had drawbacks, such as the inability to handle non-linear data. Therefore, another method for anticipating short-term loads is prepared in this research using artificial neural networks (ANN) to understand the connection between loads and weather. The neural network was trained, tested, and validated, and the predicted loads were produced by using the temperatures and historical data for peak loads in 2023 were represented by the highest electrical load for one day of the month and the temperature data per hour. The ANN model was designed for short-term load forecasting and implemented using the MATLAB package. The results showed negligible mean square error (MSE) and exceptional realism and accuracy in determining future forecast values for this type of non-linear relationship compared to other forecasting methods.