Research on Short-term Electricity Demand Forecasting Based on Deep Learning Algorithms

This paper introduces an advanced short-term electricity demand forecasting model based on Recurrent Neural Networks (RNN) combined with Long Short-Term Memory (LSTM) units. This hybrid model effectively captures the temporal dynamics in electricity consumption, incorporating features such as past energy use, temperature, humidity, and calendar variables. By utilizing a rolling prediction technique, the model addresses long-term prediction challenges and enhances accuracy. Extensive model evaluation shows a marked reduction in error metrics, including Mean Absolute Error and Root Mean Squared Error. The proposed method demonstrates significant potential for real-time energy demand forecasting, supporting grid optimization and energy management. This approach highlights the model's applicability in smart grid systems, providing more reliable predictions crucial for operational efficiency and sustainability.