A Reinforcement Learning-Based Cloud-Edge Security Strategy for abnormal maintenance in Power Distribution Network

The intelligent transformation of power distribution networks has catalyzed the application of cloud-edge-terminal collaborative architectures. However, the exponential surge in data transmission and information processing tasks across maintenance devices and edge terminals, constrained by communication bottlenecks, limited computational resources, and energy supply, creates inherent trade-offs between operational efficiency and security assurance in grid maintenance. For this reason, this paper proposes an optimization method of cloud-edge collaborative security strategies for distribution network operation and maintenance with the participation of field device. This method, on the premise of ensuring that the security protection level meets the requirements, carries out mathematical modeling and test evaluation on the communication, computing resources and energy consumption of different operation and maintenance tasks. Subsequently, quantitative evaluation and optimization are conducted for typical maintenance task based on metrics such as attribution probability and energy efficiency factors. Experimental results validate the proposed method’s effective-ness in significantly enhancing on-site operation and maintenance and cloud-edge collaborative security, providing valuable insights for relevant engineering applications.