A Comprehensive Review of Deep Residual Networks for Short-term Load Forecasting

Short-Term Load Forecasting (STLF) is a critical component of modern power systems, enabling efficient grid operation and energy management. Recent advancements in deep learning have positioned Deep Residual Network (DRN) as a promising approach for STLF, owing to their ability to capture complex and nonlinear load patterns. This paper provides a comprehensive review of DRN-based models for STLF, offering novel insights into their strengths, limitations, and future research directions. Unlike previous reviews, this work systematically evaluates DRN variants, highlighting challenges such as activation function selection, long-term dependency modelling the integration of diverse meteorological variables. Furthermore, this review proposes actionable research directions, including systematic activation function analysis, enhanced sequential modelling techniques, and multi-variable integration, to address current limitations. By bridging these gaps, this paper aims to support the development of more accurate and adaptable forecasting models, contributing to the advancement of intelligent energy management systems.