TempFocusNet: Enhancing Short-Term Electric Load Forecasting with LSTM and Temporal Attention

Precise power load forecasting is essential for efficient Smart Grid management, although it poses difficulties due to the nonlinear characteristics of electricity use. Although deep learning has demonstrated potential across multiple domains, its utilization in load forecasting requires a thorough assessment. This study fills the gap by doing a comprehensive evaluation of various deep learning architectures for short-term electric load forecasting, utilizing actual energy consumption data. We present TempFocusNet, a hybrid model that integrates Long Short-Term Memory (LSTM) dependency learning with the lightweight temporal attention strategy to improve prediction accuracy. Therefore, our methodology encompasses renewable energy consumption with instantaneous decision-making. Temp-FocusNet shows a 2.13% increase in R<sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup>, a 99.92% decrease in RMSE, a 45% improvement in MAPE, and a 91.74% decrease in MAE compared to ARIMA, the next best-performing model in terms of R<sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup>.