摘要
调度工作是电网领域的核心业务之一,随着近年来电网信息系统建设,呈现出数据多元结构复杂的特点,为满足高效响应的调度需求,文章提出了一种基于迁移学习的电网领域实体识别技术,研究了基于知识图谱的电网故障处置应用。首先,提出了一种基于迁移学习的电网故障处置文本数据的实体识别技术,该技术可以在少量标注的情况下实现较高准确度的实体识别工作,有效地解决了电网领域小样本数据实体识别准确度较低的问题。其次,构建了电网故障处置知识图谱,该图谱将电网调度领域多源异构数据转化为知识,可实现对多故障类型的决策支撑。最后,基于所建立的电网故障处置知识图谱,实现了故障处置实时辅助决策功能,并嵌套进省地协同故障处置引擎应用中。实际应用表明,通过图数据库进行关系推理得到的知识经验,可为调度人员提供有效的故障处置参考方案。
Dispatching is one of the core businesses in the field of power grid.With the construction of power grid information system in recent years,it presents the characteristics of multiple data structures and complex structures.In order to meet the dispatching requirements of efficient response,this paper proposes an entity recognition technology in the field of power grid based on migration learning,and studies the application of power grid fault handling based on knowledge graph.Firstly,an entity recognition technology based on transfer learning for power grid fault handling text data is proposed.This technology can realize entity recognition with high accuracy in the case of a small number of annotations,and effectively solve the problem of low accuracy in entity recognition of small sample data in power grid field.Secondly,the power grid fault handling knowledge graph is constructed,which converts the multi-source heterogeneous data in the power grid dispatching field into knowledge,and can realize the decision support for multiple fault types.Finally,based on the established grid fault handling knowledge graph,the real-time auxiliary decision function of fault handling is realized and nested into the application of provincial collaborative fault handling engine.Practical application shows that the knowledge and experience obtained through graph database can provide effective reference scheme for dispatchers.
作者
张锐
刘剑青
张伯远
李金星
高天露
张俊
ZHANG Rui;LIU Jianqing;ZHANG Boyuan;LI Jinxing;GAO Tianlu;ZHANG Jun(State Grid Jibei Electric Power Co.,Ltd.,Beijing 100053,China;China Electric Power Research Institute,Beijing 100192,China;School of Electrical and Automation,Wuhan University,Wuhan 430072,China)
出处
《电力信息与通信技术》
2022年第6期24-34,共11页
Electric Power Information and Communication Technology
基金
国家电网有限公司总部科技项目资助“基于泛在信息的调控保电智能技术研究及应用”(52010119006T)。
关键词
知识图谱
多源异构
智能调度
迁移学习
故障处置
实时辅助决策
knowledge graph
multi-source heterogeneous
intelligent dispatching
transfer learning
fault handling
real-time auxiliary decision
