电力动态元数据检索的大模型查询方法、装置及电子设备
By using a dynamic metadata retrieval method for electricity, the problems of low query efficiency and context overrun in power business using large language models are solved, achieving second-level response and accurate SQL generation, which is suitable for power grid equipment management and fault diagnosis.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- SICHUAN ENERGY INTERNET RES INST TSINGHUA UNIV
- Filing Date
- 2026-05-14
- Publication Date
- 2026-07-17
AI Technical Summary
Existing technologies for querying metadata using large language models in the power industry suffer from low inference efficiency and query errors caused by exceeding context length limits, failing to meet the real-time requirements of the power industry.
The method of dynamic power metadata retrieval is adopted. It performs legality verification and terminology standardization by receiving users' natural language questions, filters candidate data tables, dynamically obtains field definitions and sample data, generates structured input instructions to generate SQL statements, and performs syntax verification and security audit.
It significantly improves query efficiency, avoids the problem of exceeding the context length limit, improves query accuracy and reliability, and has good scalability and generalization ability, making it suitable for power grid equipment ledger management and fault diagnosis analysis.
Smart Images
Figure CN122220385B_ABST