基于知识图谱与大语言模型的电力行业知识检索分析系统研发与应用

doi:10.11930/j.issn.1004-9649.202409084

中国电力 ›› 2024, Vol. 57 ›› Issue (12): 198-205.DOI: 10.11930/j.issn.1004-9649.202409084

基于知识图谱与大语言模型的电力行业知识检索分析系统研发与应用

张金营¹(), 王哲峰¹(), 谢华²(), 么长英²(), 闵艳丽³(), 王新颖⁴

1. 国电电力发展股份有限公司，辽宁鞍山 100101
2. 国能信控互联技术有限公司，北京　100080
3. 同方知网数字出版技术股份有限公司，北京　100192
4. 华北电力大学电气与电子工程学院，北京　102206

收稿日期:2024-09-23 出版日期:2024-12-28 发布日期:2024-12-27
作者简介:张金营（1984—），男，高级工程师，博士，从事人工智能、分散控制系统研究，E-mail：jinying.zhang@chnenergy.com.cn
王哲峰（1977—），男，高级工程师，从事电力系统及其自动化研究，E-mail：zhefeng.wang.a@chnenergy.com.cn
谢华（1980—），男，高级工程师，研究生，从事电厂信息化研究，E-mail：13426067303@163.com
么长英（1985—），女，高级工程师，从事人工智能、机器学习、数据治理、电厂信息化和数字化转型研究，E-mail：yaochangying@126.com
闵艳丽（1982—），女，工程师，从事人工智能、数据治理、企业数字化转型研究，E-mail：myl4546@cnki.net
基金资助:
国家自然科学基金资助项目（62373151）。

Development and Application of a Knowledge Retrieval and Analysis System for the Power Industry Based on Knowledge Graph and Large Language Model

Jinying ZHANG¹(), Zhefeng WANG¹(), Hua XIE²(), Changying YAO²(), Yanli MIN³(), XINYing WANG⁴

1. Guodian Electric Power Development Co., Ltd., Anshan 100101, China
2. Guoneng Information Control Interconnection Technology Co., Ltd., Beijing 100080, China
3. Tongfang Knowledge Network Digital Publishing Technology Co., Ltd., Beijing 100192, China
4. School of Electrical and Electronic Engineering, North China Electric Power University, Beijing 102206, China

Received:2024-09-23 Online:2024-12-28 Published:2024-12-27
Supported by:
This work is supported by National Natural Science Foundation of China (No.62373151).

摘要/Abstract

摘要：

随着人工智能技术飞速发展，电力行业知识检索系统面临着技术的更新迭代。提出了一种基于知识图谱与大语言模型的电力行业知识检索分析系统。首先，借助大语言模型挖掘用户需求并理解用户的意图；然后，针对不同结构的知识信息，通过知识建模、知识抽取、知识融合等策略来构建结构化的知识图谱；最后，利用大语言模型根据用户请求和从知识子图中获取的专业知识，并将生成内容可视化展示给用户，为电力行业知识检索系统提供了新思路。

关键词: 大语言模型, 知识图谱, 知识抽取

Abstract:

With the rapid development of artificial intelligence technology, knowledge retrieval systems in the power industry are facing technological updates and iterations. A knowledge retrieval and analysis system for the power industry based on knowledge graph and big language model has been proposed. Firstly, using big language models to mine user needs and understand their intentions; Then, for knowledge information with different structures, structured knowledge graphs are constructed through strategies such as knowledge modeling, knowledge extraction, and knowledge fusion; Finally, utilizing a large language model based on user requests and professional knowledge obtained from knowledge subgraphs, and visualizing the generated content for display to users, provides a new approach for knowledge retrieval systems in the power industry.

Key words: large language model, knowledge graph, knowledge extraction

张金营, 王哲峰, 谢华, 么长英, 闵艳丽, 王新颖. 基于知识图谱与大语言模型的电力行业知识检索分析系统研发与应用[J]. 中国电力, 2024, 57(12): 198-205.

Jinying ZHANG, Zhefeng WANG, Hua XIE, Changying YAO, Yanli MIN, XINYing WANG. Development and Application of a Knowledge Retrieval and Analysis System for the Power Industry Based on Knowledge Graph and Large Language Model[J]. Electric Power, 2024, 57(12): 198-205.

导出引用管理器 EndNote|Ris|BibTeX

链接本文: https://www.electricpower.com.cn/CN/10.11930/j.issn.1004-9649.202409084

https://www.electricpower.com.cn/CN/Y2024/V57/I12/198

图/表 11

图 1 基于知识图谱与LLM的电力知识搜索流程示例

Fig.1 Example of power knowledge search process based on knowledge graph and LLM

图 2 电力知识图谱构建框架

Fig.2 The framework of knowledge graph in the power industry

图 3 知识建模流程

Fig.3 The process of knowledge modeling

图 4 知识加工流程

Fig.4 The process of knowledge processing

图 5 知识更新流程

Fig.5 The process of knowledge update

图 6 召回策略流程

Fig.6 The process of recall strategy

表 1 训练集、测试集样本构建

Table 1 Composition of training and testing set samples

任务	条数/条	备注
训练集	2000	随机采集各电力指标不同难度下的数据
	1000	开源中文数据集
	1000	开源英文数据集
	1000	开源数学推理数据集
验证集	300	简单问题、中等问题与推理问题各随采100条
测试集	300	简单问题、中等问题与推理问题各随采100条
系统测试集	300	简单问题、中等问题与推理问题各随采100条

表 2 软硬件环境配置

Table 2 Configuration of software and hardware environment

序号	配置名称	配置详情
1	中央处理器	Intel(R) Xeon(R) Silver 4110 CPU @ 2.10 GHz x86_64
2	内存	219 G
3	显卡	NVIDIA A800 80 GB PCIe 2块
4	操作系统	Centos 7
5	开发框架	Python3.8+Pytorch2.1+py2 neo2021.2.4
6	大语言模型	chatglm3
7	图谱	Neo4 j 3.5

图 7 系统专家评估

Fig.7 Expert evaluation of the system

图 8 电力行业知识检索分析系统检索界面

Fig.8 Interface for knowledge retrieval and analysis system in the power industry

图 9 检索结果

Fig.9 Search results

参考文献 29

1	田雪涵, 董坤, 赵剑锋, 等. 基于增强优化预训练语言模型的电力数据实体识别方法[J]. 智慧电力, 2024, 52 (6): 100- 107. DOI
	TIAN Xuehan, DONG Kun, ZHAO Jianfeng, et al. Entity recognition method for power data based on enhanced optimization pre-trained language model[J]. Smart Power, 2024, 52 (6): 100- 107. DOI
2	刘东, 张越, 皮俊波, 等. 面向电网调控信息智能检索的知识图谱构建及应用[J]. 中国电力, 2023, 56 (7): 78- 84.
	LIU Dong, ZHANG Yue, PI Junbo, et al. Construction and application of knowledge graph for intelligent retrieval of power grid dispatching and control information[J]. Electric Power, 2023, 56 (7): 78- 84.
3	刘中硕, 郑少明, 陶畅, 等. 继电保护装置缺陷文本专业词典构建及其语言特性分析[J]. 中国电力, 2023, 56 (7): 146- 155.
	LIU Zhongshuo, ZHENG Shaoming, TAO Chang, et al. The construction of the professional dictionary of relay protection defect text in a regional power grid and its natural language characteristics analysis[J]. Electric Power, 2023, 56 (7): 146- 155.
4	田波, 张越, 蒙飞, 等. 电网故障处置信息自适应理解框架及关键技术[J]. 中国电力, 2024, 57 (7): 188- 195.
	TIAN Bo, ZHANG Yue, MENG Fei, et al. Adaptive understanding framework and key technology of power grid fault disposal information[J]. Electric Power, 2024, 57 (7): 188- 195.
5	李莉, 时榕良, 郭旭, 等. 融合大模型与图神经网络的电力设备缺陷诊断[J]. 计算机科学与探索, 2024, 18 (10): 2643- 2655. DOI
	LI Li, SHI Rongliang, GUO Xu, et al. Diagnosis of power system defects by large language models and graph neural networks[J]. Journal of Frontiers of Computer Science and Technology, 2024, 18 (10): 2643- 2655. DOI
6	邹国惠, 魏嘉隆, 王超, 等. 基于知识图谱的变电设备画像技术[J]. 广东电力, 2024, 37 (1): 86- 93.
	ZOU Guohui, WEI Jialong, WANG Chao, et al. Knowledge graph-based portrait depiction technology for substation equipment[J]. Guangdong Electric Power, 2024, 37 (1): 86- 93.
7	陈志忠, 熊泽森, 姚东, 等. 基于知识图谱的电力杆塔主要构件识别方法研究[J]. 浙江电力, 2024, 43 (5): 100- 108.
	CHEN Zhizhong, XIONG Zesen, YAO Dong, et al. Research on a recognition method of main components of electric power towers using knowledge graph[J]. Zhejiang Electric Power, 2024, 43 (5): 100- 108.
8	陈将宏, 饶佳黎, 李伟亮, 等. 基于向量自回归模型的电网虚假数据注入攻击检测[J]. 电力科学与技术学报, 2024, 39 (3): 1- 9.
	CHEN Jianghong, RAO Jiali, LI Weiliang, et al. Detection method of false data injection attacks on power grids based on vector auto-regression model[J]. Journal of Electric Power Science and Technology, 2024, 39 (3): 1- 9.
9	周义棋, 刘畅, 龙增, 等. 电网应急预案知识图谱构建方法与应用[J]. 中国安全生产科学技术, 2023, 19 (1): 5- 13.
	ZHOU Yiqi, LIU Chang, LONG Zeng, et al. Construction method and application of knowledge graph in emergency plans for power grid[J]. Journal of Safety Science and Technology, 2023, 19 (1): 5- 13.
10	蒲天骄, 谈元鹏, 彭国政, 等. 电力领域知识图谱的构建与应用[J]. 电网技术, 2021, 45 (6): 2080- 2091.
	PU Tianjiao, TAN Yuanpeng, PENG Guozheng, et al. Construction and application of knowledge graph in the electric power field[J]. Power System Technology, 2021, 45 (6): 2080- 2091.
11	殷婕. 基于数据挖掘的输电断面动态极限传输功率在线构建方法[J]. 东北电力大学学报, 2023, 43 (1): 69- 75.
	YIN Jie. Online construction of dynamic total transfer capability of transmission interface based onData mining technology[J]. Journal of Northeast Electric Power University, 2023, 43 (1): 69- 75.
12	李彦儒, 王慧芳, 陈昊, 等. 电力设备缺陷文本质量保证与评级软件的开发及应用[J]. 浙江电力, 2022, 41 (7): 78- 85.
	LI Yanru, WANG Huifang, CHEN Hao, et al. Development and application of quality assurance and rating software program for power equipment defect texts[J]. Zhejiang Electric Power, 2022, 41 (7): 78- 85.
13	石宏岩, 李艳来. 基于知识图谱对吸收式热泵的研究概况与趋势预测[J]. 内蒙古电力技术, 2023, 41 (4): 32- 40.
	SHI Hongyan, LI Yanlai. Research overview and trend prediction of absorption heat pumps based on knowledge graph[J]. Inner Mongolia Electric Power, 2023, 41 (4): 32- 40.
14	张兰. 电网企业标准数字化发展现状及趋势分析[J]. 湖南电力, 2023, 43 (1): 95- 98, 102. DOI
	ZHANG Lan. Development status and trend analysis of standard digitization in power grid enterprise[J]. Hunan Electric Power, 2023, 43 (1): 95- 98, 102. DOI
15	LIU X, JI K X, FU Y C, et al. P-tuning v2: prompt tuning can be comparable to fine-tuning universally across scales and tasks[EB/OL]. (2021-11-12)[2024-09-22]. https://arxiv.org/abs/2110.07602v3.
16	周园春, 王卫军, 乔子越, 等. 科技大数据知识图谱构建方法及应用研究综述[J]. 中国科学: 信息科学, 2020, 50 (7): 957- 987. DOI
	ZHOU Yuanchun, WANG Weijun, QIAO Ziyue, et al. A survey on the construction methods and applications of sci-tech big data knowledge graph[J]. Scientia Sinica (Informationis), 2020, 50 (7): 957- 987. DOI
17	石水倩, 金晶, 沈耕宇, 等. 基于多元相似度融合的中文命名实体消歧方法[J]. 数据分析与知识发现, 2024, 8 (2): 56- 64.
	SHI Shuiqian, JIN Jing, SHEN Gengyu, et al. Chinese named entity disambiguation based on multivariate similarity fusion[J]. Data Analysis and Knowledge Discovery, 2024, 8 (2): 56- 64.
18	JOSHI M, CHEN D Q, LIU Y H, et al. SpanBERT: improving pre-training by representing and predicting spans[EB/OL]. (2019-12-12)[2024-08-15]. https://arxiv.org/abs/1907.10529v3.
19	DOBROVOLSKII V. Word-level coreference resolution[EB/OL]. (2021-11-10)[2024-08-19]. https://arxiv.org/abs/2109.04127v1.
20	GANEA O E, HOFMANN T. Deep joint entity disambiguation with local neural attention[EB/OL]. (2017-10-12)[2024-07-16]. https://arxiv.org/abs/1704.04920v3.
21	张鹤译, 王鑫, 韩立帆, 等. 大语言模型融合知识图谱的问答系统研究[J]. 计算机科学与探索, 2023, 17 (10): 2377- 2388. DOI
	ZHANG Heyi, WANG Xin, HAN Lifan, et al. Research on question answering system on joint of knowledge graph and large language models[J]. Journal of Frontiers of Computer Science and Technology, 2023, 17 (10): 2377- 2388. DOI
22	刘洪波, 盖雪扬, 孙黎, 等. 计及数据降维和数据清洗的超短期风电功率预测[J]. 东北电力大学学报, 2023, 43 (4): 1- 9.
	LIU Hongbo, GAI Xueyang, SUN Li, et al. Ultra-short-term wind power prediction considering data reduction and data cleaning[J]. Journal of Northeast Electric Power University, 2023, 43 (4): 1- 9.
23	徐恒辉, 姚杰, 周萍, 等. 基于运行数据的光伏电站状态评估方法研究[J]. 电力科技与环保, 2023, 39 (5): 450- 456.
	XU Henghui, YAO Jie, ZHOU Ping, et al. Research on photovoltaic power plant state evaluation based on operating data[J]. Electric Power Technology and Environmental Protection, 2023, 39 (5): 450- 456.
24	吕耀棠, 李志中, 吴小刚. 基于大数据技术的电力调度准时性全流程管控系统设计与应用[J]. 广东电力, 2023, 36 (12): 76- 82. DOI
	LV Yaotang, LI Zhizhong, WU Xiaogang. Design and application of punctual whole-process control system of power dispatch based on big data technology[J]. Guangdong Electric Power, 2023, 36 (12): 76- 82. DOI
25	彭鹏, 薛东. 基于边缘代理的云边协同物联网框架设计及其应用研究[J]. 广东电力, 2023, 36 (5): 18- 26. DOI
	PENG Peng, XUE Dong. Research on design of cloud side collaborative IOT framework based on edge agent and its application[J]. Guangdong Electric Power, 2023, 36 (5): 18- 26. DOI
26	陈祖歌, 毛冬, 饶涵宇, 等. 面向物联网云计算的能源大数据存储优化算法[J]. 浙江电力, 2023, 42 (8): 19- 26.
	CHEN Zuge, MAO Dong, RAO Hanyu, et al. Optimization algorithm of energy big data storage for cloud computing in Internet of Things[J]. Zhejiang Electric Power, 2023, 42 (8): 19- 26.
27	刘栋, 张广军, 袁凌, 等. 风力发电机组叶片二维翼型气动噪声数值模拟[J]. 电力科技与环保, 2022, 38 (3): 166- 174.
	LIU Dong, ZHANG Guangjun, YUAN Ling, et al. Numerical simulation of aerodynamic noise of two-dimensional airfoil of wind turbine blades[J]. Electric Power Technology and Environmental Protection, 2022, 38 (3): 166- 174.
28	李邦源, 李昂, 石恒初, 等. 计及分布式电源输出特性的主动配电网合环电流计算方法[J]. 电网与清洁能源, 2023, 39 (6): 115- 123.
	LI Bangyuan, LI Ang, SHI Hengchu, et al. A calculation method of the closing loop current in active distribution networks with distributed power output characteristics considered[J]. Power System and Clean Energy, 2023, 39 (6): 115- 123.
29	许训炜, 沈希澄, 周霞, 等. 基于数据驱动的源网荷储协同控制系统网络攻击关联性分析[J]. 浙江电力, 2023, 42 (2): 76- 82.
	XU Xunwei, SHEN Xicheng, ZHOU Xia, et al. A data-driven correlation analysis of cyberattack on coordinated source-networkload-storage control system[J]. Zhejiang Electric Power, 2023, 42 (2): 76- 82.

[1]	田波, 张越, 蒙飞, 单连飞, 高海洋, 田坤, 乔咏田. 电网故障处置信息自适应理解框架及关键技术[J]. 中国电力, 2024, 57(7): 188-195.
[2]	刘东, 张越, 皮俊波, 单连飞, 刘赫, 宋鹏程, 姜涛. 面向电网调控信息智能检索的知识图谱构建及应用[J]. 中国电力, 2023, 56(7): 78-84.
[3]	梅冰笑, 周金辉, 孙翔. 基于知识图谱与细胞自动机模型的配电网信息系统风险分析[J]. 中国电力, 2022, 55(10): 23-31.
[4]	李晓露, 左璇, 刘日亮, 陆一鸣, 李聪利, 林顺富. 基于形状约束语言的电网模型知识图谱验证方法[J]. 中国电力, 2022, 55(1): 119-125,228.
[5]	李新鹏, 徐建航, 郭子明, 李军良, 宁文元, 王震学. 调度自动化系统知识图谱的构建与应用[J]. 中国电力, 2019, 52(2): 70-77,157.

基于知识图谱与大语言模型的电力行业知识检索分析系统研发与应用

Development and Application of a Knowledge Retrieval and Analysis System for the Power Industry Based on Knowledge Graph and Large Language Model

RichHTML

PDF (PC)

可视化

摘要/Abstract

引用本文

使用本文

图/表 11

参考文献 29

相关文章 5

编辑推荐

Metrics

模态框（Modal）标题

基于知识图谱与大语言模型的电力行业知识检索分析系统研发与应用

Development and Application of a Knowledge Retrieval and Analysis System for the Power Industry Based on Knowledge Graph and Large Language Model

RichHTML

PDF (PC)

可视化

摘要/Abstract

引用本文

使用本文

图/表 11

参考文献 29

相关文章 5

编辑推荐

Metrics