Short-term load forecasting using LSTM

One of the most crucial elements for power system operators is short-term load forecasting (STLF), enabling efficient resource allocation and grid stability. Accurate short-term load forecasting predicts the electrical load in small time horizons, that start from hours up to days with data based on history and real-time data. This project utilises LSTM networks in predicting power load at Godishala Substation by having features that include voltage, current, temperature, and humidity besides the time-related attributes. The results are compared using mean absolute error and root mean squared error. The outcomes show that the LSTM model outperforms GRU throughout the experiment and delivers enhanced predictive accuracy. GUI is integrated into the system to allow users to input data directly for real-time load forecasts, hence improving the practical usability of the system. This study supports the LSTM model’s efficacy for STLF decision-making in operational tasks.