Outage-Prone Neural Networks Analysis for Short-Term Electrical Demand Forecasting

Short-term load forecasting (STLF) is crucial for efficient power distribution, generation, and utilization, particularly in developing countries like Pakistan, where demandsupply gaps are substantial. Summer electricity load patterns exhibit high nonlinearity due to frequent, intense, and unplanned power outages, resulting in discontinuities in consumption and complicating forecasting and planning. This study explored the incorporation of power outage data as an explanatory variable in forecasting models. Initially, a statistical regression model was tested for its simplicity and robustness, but it showed limited accuracy. Various artificial neural network (ANN) models were trained and tested to capture complex nonlinear relationships more effectively. A comparative analysis between the regression and ANN models used several error metrics. The results show that outage-informed Feedforward Backpropagation (FFBP) Neural Networks achieved the lowest Mean Absolute Percentage Error (MAPE) of <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">1. 3 %</tex> and Mean Square Error (MSE) of 7 %, reducing errors by up to 98 % compared to traditional forecasting and other ANN architectures. This study highlights the significant benefits of outage-aware neural networks in improving short-term electrical load forecasting in outage-prone environments.