基于知识图谱辅助大模型的多源电力系统场景问答技术

打开文本图片集

中图分类号：TM74;TQ018 文献标识码：A 文章编号：1001-5922（2025）10-0191-04

Multi-source power system scenario question answering technology based on knowledge graph assisted large model

DUJianguang，CHENZhenyu，ZHU Tianyou，LIJiwei，YANGShiyu （StateGrid Corporationof ChinaBig Data Center，Beijing1OOO52，China）

Abstract： The multi-source power system involves a wide range of data sources and diverse formats. These data need to be processd in a complex manner during the fusion process，increasing the time cost of data processing， and thus affecting the response speed of the question answering system.Therefore，this paper focuses on the multi-source power system scenario question answering technology based on knowledge graph-assisted large model， aiming to improvethe inteligent level of power system management.The effective fusion of multi-source power systemdatais realized to ensure thecomprehensivenessand accuracyofthedata.Using knowledge extraction technology to extract key information from massive dataand construct a structured knowledge system;through the finely constructed knowledge graph to asist thelarge model，deepen the understanding of thecomplexrelationshipofthe power system; using knowledge graph template matching question answering to realize natural language procesing and accurate answer return of user queries，the information interaction eficiency and decision support abilityof power system are improved.The experimental results significantly improve the information processing eficiency and response speed，andverifyitsgreatpotential inintegrating multi-source data，quicklyunderstandingandresponding to complex scenarios of power systems.

Key words： knowledge graph; knowledge graph asisted large model; multi source power system scenario; scene Q&A; scene Q&A technology

基于知识图谱辅助大模型的多源电力系统场景问答技术，为电力系统的智能化转型开辟了一条新路径。（剩余5639字）