Research on Intelligent Dispatching Method of Power System Based on Chain of Thought and Knowledge Graph of Large Language Model

With the rapid expansion of new energy installations and power grid scale, traditional manual dispatching methods can no longer meet increasingly complex system requirements. This paper presents an intelligent dispatching methodology that integrates large language model chain-of-thought reasoning with knowledge graph technology. We propose a framework that simulates professional regulators' decision-making processes through chain-of-thought reasoning, enabling comprehensive understanding of complex dispatching scenarios. A formalized knowledge representation is constructed using knowledge graph technology to provide reliable support for the reasoning process. Additionally, we introduce an innovative dispatching strategy generation method that transforms professional expertise into executable dispatching schemes through prompt engineering and context management mechanisms. Experimental results demonstrate that this method achieves superior performance in complex dispatching scenarios, maintaining high accuracy while significantly reducing response time. The methodology exhibits robust stability and scalability, presenting a novel technical approach for power system intelligent dispatching by transitioning from experience-driven to intelligence-driven control. This advancement offers significant theoretical value and practical application potential.