With the continuous expansion of distribution networks and the ongoing advancement of automation levels, the variety and complexity of distribution network operation commands have rapidly increased. Traditional processing methods, primarily relying on manual parsing and rule matching, can no longer meet the demands for safe and efficient dispatch. To address issues such as semantic ambiguity, inconsistent formats, and low manual recognition efficiency in distribution network operation commands, this paper proposes an intelligent parsing and visualized interaction method driven by large models. This approach fully leverages the semantic understanding and contextual reasoning capabilities of Large Language Models (LLMs), combined with domain-specific knowledge graphs, to achieve semantic modeling, intent recognition, and structured parsing of commands. Methodologically, this paper constructs a domain-knowledge-integrated LLM parsing framework. Through fine-tuning and multi-task learning strategies, it optimizes the model's semantic adaptation capabilities. Parsing results are dynamically presented in multidimensional visualizations, enabling visual tracking of operational workflows and human-machine interaction. Prototype validation demonstrates that this approach significantly improves command parsing accuracy and response speed, reduces manual intervention, and enhances the safety and intelligence of distribution network operations. These findings provide a novel technical pathway for semantic understanding and interactive visualization in intelligent distribution network dispatch systems, holding significant theoretical implications and engineering application value for advancing the digital and intelligent transformation of distribution grids.
Li et al. (Sun,) studied this question.