摘要
针对电力公司在开展变压器设备运检过程中存在的非结构化文本数据难以利用、全口径数据难以深度融合、设备知识应用深度较浅等难题,基于语义网、知识图谱、自然语言处理等人工智能技术,对开展设备智能管理的关键技术进行了研究,提出了支撑变压器智能管理的智能技术框架,包括非结构化文本智能识别与提取、以设备为中心的设备知识表示与存储、设备知识服务应用三部分;形成了设备语义提取模型、设备语义相似度计算模型、基于深度神经网络的智能问答模型等三个智能模型;总结了该技术在变压器设备状态评价报告自动化审核、设备信息灵活查询、基于设备故障知识的辅助诊断三个场景的应用成果;提出了基于知识的设备智能技术下一步研究的方向。
During transformer-equipment operation and maintenance, power companies encounter problems such as generation of difficult-to-use unstructured text data, full-caliber data that are difficult to integrate deeply, and application of shallow equipment knowledge application. Using artificial-intelligence techniques, such as the semantic web, knowledge map, and natural language processing, this study investigates the key technologies used for intelligent servicing of equipment. Accordingly, an intelligent technology framework for transformer operation and maintenance is proposed. This framework comprises three components—intelligent unstructured-text recognition and extraction, devicecentered knowledge representation and storage, and intelligent device knowledge application. Three deep-neural-networkbased intelligent models for device semantic extraction, device semantic-similarity calculation, and intelligent question answering are proposed to summarize three scenarios—review of the transformer running-status evaluation report, flexible equipment information querying, and auxiliary diagnosis. Finally, directions for future research on knowledge-based equipment intelligence technologies are identified.
作者
张敏杰
徐宁
胡俊华
王宇飞
李晨
徐剑波
张诗玉
ZHANG Minjie;XU Ning;HU Junhua;WANG Yufei;Li Chen;XU Jianbo;ZHANG Shiyu(Kindi(Beijing)Network Technology Co.,Ltd.,Haidian District,Beijing 100089,China;Electric Power Research Institute of State Grid Zhejiang Electric Power Co.,Ltd.,Hangzhou 310014,Zhejiang Province,China;State Grid Zhejiang Electric Power Co.,Ltd.,Hangzhou 310007,Zhejiang Province,China;Beijing Institute of Big Data Research,Haidian District,Beijing 100089,China)
出处
《全球能源互联网》
2020年第6期607-617,共11页
Journal of Global Energy Interconnection
基金
国家电网公司科技项目(5500-202019090A-0-0-00)
关键词
智能运检
知识图谱
变压器故障辅助诊断
智能问答
intelligent operation and maintenance
knowledge graph
transformer fault diagnosis
intelligent question answering
