| 引用本文: | 戴宇欣,张俊,季知祥,刘明忠,高天露,郑永康,姚良忠.基于功能缺陷文本的电力系统二次设备智能诊断与辅助决策[J].电力自动化设备,2021,41(6): |
| DAI Yuxin,ZHANG Jun,JI Zhixiang,LIU Mingzhong,GAO Tianlu,ZHENG Yongkang,YAO Liangzhong.Intelligent diagnosis and auxiliary decision of power system secondary equipment based on functional defect text[J].Electric Power Automation Equipment,2021,41(6): |
|
| 摘要: |
| 利用电力系统二次设备功能缺陷文本数据，建立了基于双向长短时记忆网络与条件随机场(BiLSTM-CRF)模型的文本信息抽取模型。在此基础上，为了进一步将数据中蕴含的知识价值应用到电力系统生产、管理过程中，构建了电力系统二次设备功能缺陷知识图谱，将各类数据间所含语义信息融入各类实体间的关系约束，建立了基于BiLSTM-CRF模型与知识图谱的二次设备功能缺陷智能诊断与辅助决策平台。该平台可依据缺陷设备类型与缺陷现象快速诊断设备的缺陷部位及原因，并推荐合理的解决措施。算例分析结果表明，相较于传统的命名实体识别算法、BiLSTM-softmax以及Seq2Seq-Attention模型，所采用BiLSTM-CRF模型的精确率、召回率、F1值这3项评估指标均有较大提升，所建平台能很好地挖掘、应用电力文本数据知识与价值，为电力系统二次设备功能缺陷处理提供有益参考。 |
| 关键词: 电力系统 二次设备 信息抽取 知识应用 知识图谱 BiLSTM-CRF 智能诊断 辅助决策 |
| DOI：10.16081/j.epae.202106006 |
| 分类号:TM761 |
| 基金项目: |
|
| Intelligent diagnosis and auxiliary decision of power system secondary equipment based on functional defect text |
|
DAI Yuxin1, ZHANG Jun1, JI Zhixiang2, LIU Mingzhong3, GAO Tianlu1, ZHENG Yongkang3, YAO Liangzhong1
|
|
1.School of Electrical Engineering and Automation, Wuhan University, Wuhan 430072, China;2.China Electric Power Research Institute, Beijing 100192, China;3.State Grid Sichuan Electric Power Research Institute, Chengdu 610041, China
|
| Abstract: |
| A text information extraction model based on BiLSTM-CRF(Bi-directional Long Short-Term Memory and Conditional Random Field) model is established by using the functional defect text of power system secondary equipment. On this basis, in order to further apply the knowledge value contained in the data to the production and management process of power system, the knowledge graph for functional defects of power system secondary equipment is constructed, which can integrate the semantic information contained in various types of data into the relationship constraints among various types of entity. And an intelligent diagnosis and auxiliary decision platform for functional defects of secondary equipment based on BiLSTM-CRF model and knowledge graph is established. The platform can quickly diagnose the defective parts and causes of the equipment according to the type and phenomenon of defective equipment, and then recommend reasonable solutions. The numerical example analysis results show that, compared with the traditional named entity re-cognition algorithm, BiLSTM-Softmax model and Seq2Seq-Attention model, the evaluation indexes of accurate rate, recall rate and F1 value for BiLSTM-CRF model are greatly improved, and the established platform can well mine and apply the knowledge and value of power text data, providing a useful reference for processing the functional defects of power system secondary equipment. |
| Key words: electric power systems secondary equipment information extraction knowledge application know-ledge graph BiLSTM-CRF intelligent diagnosis auxiliary decision |
