Classification-Enhanced and hybrid forecasting paradigms in recent Short-Term electricity load Forecasting: A PRISMA-Based Systematic review

Energy Conversion and Management X, vol. 31, pp. 102065

Abstract

Accurate short-term electricity load forecasting (STLF) is essential for reliable power system operation, electricity market efficiency, demand-side management, and renewable energy integration. However, increasing distributed energy resources, electric vehicle adoption, smart-grid deployment, and changing consumer behavior have introduced nonlinear and heterogeneous load patterns that challenge conventional forecasting models. This study presents a PRISMA 2020-based systematic review of 74 Q1-ranked journal articles published between 2023 and 2025. The reviewed studies are categorized into statistical, machine learning, deep learning, and hybrid forecasting paradigms, with particular emphasis on distinguishing classification-enhanced pipelines from direct forecasting models. The synthesis shows that classification and clustering are frequently associated with improved robustness when load profiles are heterogeneous, event-driven, or behaviorally segmented, such as during holidays, weekends, regime shifts, or customer-group-specific demand patterns. However, their benefits are context-dependent and should not be interpreted as universal superiority. Direct forecasting models using decomposition, attention mechanisms, transfer learning, or Transformer-based architecture remain competitive under stable, aggregated, or well-preprocessed conditions. The review further highlights key limitations in the literature, including inconsistent validation protocols, limited risk-of-bias assessment, insufficient computational benchmarking, and underreported deployment feasibility. Overall, this study provides a structured synthesis of recent STLF research and identifies future directions toward interpretable, scalable, computationally feasible, and standardized forecasting frameworks.

Authors 5

1. Ei Thu Thu Tun

   Thammasat University

   Affiliation as printed

   School of Management Technology, Sirindhorn International Institute of Technology, Thammasat University, Pathum Thani 12000, Thailand

2. Kamal Chapagain

   Kathmandu University

   Affiliation as printed

   Department of Electrical & Electronics Engineering, Kathmandu University, Kathmandu 6250, Nepal

3. Yung-Cheol Byun

   Jeju National University

   Affiliation as printed

   Department of Computer Engineering, Head of RIS Office for High-Intelligent Service, Jeju National University, Jeju 63243, South Korea

4. Pyae Pyae Phyo Aachen Institute for Automation of Complex Power Systems

   RWTH Aachen University

   Affiliation as printed

   Institute for Automation of Complex Power Systems, RWTH Aachen University, Aachen 52074, Germany

5. Chawalit Jeenanunta corresponding

   Thammasat University

   Affiliation as printed

   School of Management Technology, Sirindhorn International Institute of Technology, Thammasat University, Pathum Thani 12000, Thailand