A Digital-Twin-Driven Cyber–Physical Framework for Real-Time Energy Management and Secure Operation of Renewable Energy Systems

Abstract

Real-time optimal operation and control acts as the intelligent core of renewable energy systems, yet intermittent generation and time-varying loads introduce uncertainties that hinder secure and efficient operation. To address this issue, this paper proposes a digital-twin-driven cyber–physical framework for real-time energy management and secure operation, enabling tight synchronization and closed-loop interaction between physical assets and their digital counterparts. A simplified multi-source data fusion Transformer is developed to improve forecasting accuracy by continuously integrating historical power data and key environmental factors (e.g., temperature, wind speed, and solar radiation). Based on the twin-enabled predictive information and state feedback, a Transformer–Adaptive Dynamic Programming (TM-ADP) method is further proposed for real-time optimal scheduling of grid-connected renewable energy systems. In addition, a YOLOv8-based multi-target detection module is incorporated to enhance operational safety through intelligent visual monitoring of photovoltaic panels. Experiments on a digital twin platform validate that the proposed framework improves operational efficiency, reliability, and safety under uncertain operating conditions.