An optimized system for predicting energy usage in smart grids using temporal fusion transformer and Aquila optimizer

Area of Science:

• Energy Systems Engineering
• Artificial Intelligence
• Computational Science

Background:

• Smart grids require accurate energy consumption forecasts due to renewable energy integration and smart meter data.
• Existing forecasting models face challenges with complex time-series data and hyperparameter optimization.

Purpose of the Study:

• To develop and evaluate an optimized system for predicting energy usage in smart grids.
• To enhance the accuracy and efficiency of energy consumption forecasting models.

Main Methods:

• Integration of the Temporal Fusion Transformer (TFT) for time-series analysis.
• Application of the Aquila Optimizer (AO) for hyperparameter tuning of the TFT model.
• Comparative analysis against traditional models like LSTM and CNN-BiLSTM.

Main Results:

• The proposed AO-TFT model demonstrated superior accuracy with a lower Root Mean Square Error (RMSE) of 0.48 and Mean Absolute Error (MAE) of 0.31.
• The AO-TFT model exhibited faster computation times compared to traditional methods.
• Analysis identified key factors influencing energy prediction, including building types, weather, and load variations.

Conclusions:

• The AO-TFT model offers a highly accurate and efficient solution for smart grid energy usage prediction.
• Optimized hyperparameter tuning is crucial for enhancing the performance of deep learning forecasting models.
• Further research into hybrid optimization and adaptive models can advance dynamic grid management.