A review of SCADA-based predictive models for distribution network load forecasting

The increasing complexity of modern power distribution networks, driven by renewable energy integration, distributed generation, and dynamic electricity demand patterns, has created a strong need for intelligent and adaptive load forecasting systems. This review examines the role of Supervisory Control and Data Acquisition (SCADA) systems in predictive load forecasting for power distribution networks, emphasizing their contribution to real-time monitoring, operational automation, and data-driven decision-making. The study evaluates conventional statistical approaches, machine learning techniques, deep learning architectures, and hybrid forecasting models used for predictive load profiling. The review highlights that machine learning-based methods generally offer superior forecasting accuracy and adaptability compared to traditional statistical models, particularly in handling nonlinear and high-dimensional power system data. Hybrid forecasting methods further improve robustness and predictive performance by integrating multiple analytical techniques and leveraging SCADA-derived operational information. However, despite significant progress, key challenges remain, including limited feeder-level forecasting, weak real-time implementation, fragmented SCADA integration, cybersecurity vulnerabilities, scalability constraints, and poor interpretability of advanced artificial intelligence models. The study identifies emerging opportunities in explainable AI, federated learning, edge computing, digital twins, transfer learning, reinforcement learning, and cyber-resilient forecasting as critical directions for future research and implementation. Overall, SCADA-based predictive load forecasting is positioned as a foundational component of next-generation smart grids, capable of improving energy efficiency, enhancing reliability, enabling adaptive control, and supporting sustainable power distribution in increasingly decentralized electricity networks.