Power Communication Planning Data Query System Based on the NL2SQL Framework

With the explosive growth of data volume and the complication of business requirements in the field of power communication planning, efficient data acquisition and analysis have become key challenges in the industry. This paper proposes a power communication planning data query system that integrates a domain-knowledge-enhanced NL2SQL framework. The system achieves accurate conversion from natural language to SQL through the Qwen2.5-7B-Instruct large language model and designs a closed-loop interaction mode of "natural language → SQL → natural language response". The system architecture covers the user interaction layer, large model processing layer, data storage layer, and result generation layer, innovatively introducing domain knowledge prompt templates, intelligent SQL execution and error handling mechanisms, and multi-level cache optimization strategies. Experimental results show that the Elec-TS model (for text-to-SQL) proposed in this paper achieves an accuracy rate of 91.40%, a recall rate of 100%, and an F1 value of 95.51% on a professional dataset of 1,000 entries. The business fit of the PC-RG model (for result generation) reaches 4.7/5, significantly improving data query efficiency and user experience in power communication planning scenarios and providing a new technical path for industry digital transformation.