| 吴子辰,顾彬,张云翔,王莘然,高华.基于大模型和知识图谱的电网检修计划智能编排[J].情报工程,2025,11(1):029-041 |
| 基于大模型和知识图谱的电网检修计划智能编排 |
| Intelligent Orchestration of Grid Maintenance Plan Based on Large Model and Knowledge Graph |
| |
| DOI： |
| 中文关键词: 大语言模型；知识图谱；智能编排；实体识别 |
| 英文关键词: Large Language Model; Knowledge Graph; Intelligent Orchestration; Entity Recognition |
| 基金项目:江苏省电力有限公司项目“基于数字孪生与自然语言处理的电力通信检修辅助决策关键技术研究”（J2023108）。 |
| 作者 | 单位 | | 吴子辰 | 国网江苏省电力有限公司信息通信分公司 南京 210024 | | 顾彬 | 国网江苏省电力有限公司信息通信分公司 南京 210024 | | 张云翔 | 国网江苏省电力有限公司信息通信分公司 南京 210024 | | 王莘然 | 国网江苏省电力有限公司信息通信分公司 南京 210024 | | 高华 | 国网江苏省电力有限公司信息通信分公司 南京 210024 |
|
| 摘要点击次数: 45 |
| 全文下载次数: 13 |
| 中文摘要: |
| [目的/意义]提出一种结合大语言模型和知识图谱技术的电网检修计划智能编排方法，以提高检修计划编排的效率和质量。[方法/过程]利用大语言模型处理自然语言查询，并结合智能编排算法和图数据库技术，从电网公司诸多论文和专利中提取检修计划相关数据与总结编排规则，构建大规模问答数据集进行训练和验证。通过引入知识图谱技术，基于检修票据记录构建检修计划知识图谱，本方法能够更有效地表示和推理检修计划中的实体关系，从而提高实体识别和智能编排任务的性能。[ 结果/ 结论] 主要研究结论显示，相较于现有技术，所提方法在实体识别和智能编排任务中性能更优，准确率更高，且能显著减少人力成本投入，为电网检修计划智能化提供了有效解决方案。 |
| 英文摘要: |
| [Objective/Significance] This paper aims to propose an intelligent orchestration method for Grid Maintenance Plan that integrates large language models with knowledge graph technology to enhance the efficiency and quality of maintenance plan scheduling. [Methods/Processes] The research methodology involves using large language models to process natural language queries, combined with intelligent orchestration algorithms and graph database technology. This approach extracts data related to maintenance plans and summarizes orchestration rules from numerous patents of the power grid company to construct a largescale question-answering dataset for training and validation. By introducing knowledge graph technology, this method constructs a maintenance plan knowledge graph based on historical maintenance ticket records, enabling more effective representation and reasoning of entity relationships within the maintenance plan, thereby improving the performance of entity recognition and intelligent orchestration tasks. [Results/Conclusions] The main research findings indicate that compared to existing technologies, the proposed method performs better in entity recognition and intelligent orchestration tasks, with higher accuracy, and significantly reduces the investment of human labor costs, providing an effective solution for the intelligent orchestration of grid maintenance plans. |
|
查看全文
查看/发表评论 下载PDF阅读器 |
| 关闭 |
