LLM-Enhanced PSO for ECU Configuration in Wireless-Supported Distribution Network Self-Restoration

Rapid fault self-restoration in complex power distribution networks is crucial. Edge Computing Units (ECUs) offer decentralized control, but their optimal configuration is challenging. This paper proposes a Large Language Model (LLM) enhanced Particle Swarm Optimization (PSO) framework for ECU configuration, explicitly considering advanced wireless communication (e.g., 5G/6G, LoRa) characteristics. The LLM aids in intelligent population initialization and adaptive particle guidance within PSO. This approach aims to optimize ECU placement and inter-ECU data exchange. Simulations on IEEE test systems show that the LLM-enhanced PSO significantly improves ECU configurations, reduces communication delays, and enhances self-restoration performance, thereby bolstering smart grid resilience.