Energy Demand Forecasting for High Energy Consumers: A Case Study in Brazil

Energy demand is a variable in constant growth, influenced by economic, geographic, climatic, and other factors. Accurate forecasting of energy demand is crucial for developing electrical systems. It is particularly interesting to energy-generating plants, distributors, and final consumers aiming for optimized energy purchases. This study presents the results of long-term energy demand forecasting for a high-energy consumer in Brazil using Long Short-Term Memory (LSTM) neural networks. A database covering the period from 2017 to 2024 was used, and the forecasted results were compared with actual energy bills. The findings demonstrate the accuracy of the LSTM technique, though it shows weaknesses when faced with unforeseen events.