郑俊翔,施正钗,周泰斌,陆千毅,黄达铁
ZHENG Junxiang,SHI Zhengchai,ZHOU Taibin,LU Qianyi,HUANG Datie

• 1:国网浙江省电力有限公司温州供电公司

摘要(Abstract)：

为实现对SCADA(数据采集与监视)系统海量告警信息的缺陷辨识和风险分析,提出了一种基于自然语义分析的电网告警信息文本缺陷风险预警方法。将基于BiLSTM-Attention神经网络的语义分析技术与模糊化缺陷风险评估方法相结合,首先对告警信息文本进行数据预处理;利用Word2vec模型进行词嵌入向量表征后作为BiLSTM层输入;然后通过注意力机制增强对告警信息中与缺陷程度相关的特征并完成缺陷等级分类;最后利用缺陷评估方案进行N-1风险定级和预警。经测试和应用分析表明:基于该方法的判断模型能够精准完成告警信息的缺陷分类定级,实现告警信息的缺陷风险预警,可为调控人员缺陷处置提供辅助决策。
For defect identification and risk analysis on enormous alarm information of SCADA(supervisory control and data acquisition), the paper proposes an early warning method of power grid alarm information defect risk based on semantic analysis: combining the fuzzy defect risk assessment method with semantic analysis technique based on the BiLSTM-Attention to preprocesses the alarm information texts; then the Word2 vec model is used for word embedding vectors representation and then used as the input of BiLSTM layer; the defect level feature representation in the alarm information was enhanced through attention mechanism, and the defect level classification of alarm information was realized. Finally, the defect assessment scheme was used for N-1 risk rating and warning. The testing and application analysis results demonstrate that the diagnosis model based on this method can accurately classify and rate the defect of the alarm information, and can realize the defect risk warning of the alarm information, providing decision-making support for dispatch personnel to deal with equipment defects.

关键词(KeyWords)： 告警信息;缺陷分类;注意力机制;双向长短时记忆网络;风险预警
alarm information;defect classification;attention mechanism;BiLSTM;risk warning

Abstract:

Keywords:

基金项目(Foundation): 国网浙江省电力有限公司科技项目(5211WZ18007F)

作者(Author): 郑俊翔,施正钗,周泰斌,陆千毅,黄达铁
ZHENG Junxiang,SHI Zhengchai,ZHOU Taibin,LU Qianyi,HUANG Datie

DOI: 10.19585/j.zjdl.202108005

参考文献(References)：

• [1]尹积军,夏清.能源互联网形态下多元融合高弹性电网的概念设计与探索[J].中国电机工程学报,2021,41(2):486-497.
• [2]李明节,陶洪铸,许洪强,等.电网调控领域人工智能技术框架与应用展望[J].电网技术,2020,44(2):393-400.
• [3]周念成,廖建权,王强钢,等.深度学习在智能电网中的应用现状分析和展望[J].电力系统自动化,2019,43(4):180-197.
• [4]曹靖,陈陆燊,邱剑,等．基于语义框架的电网缺陷文本挖掘技术及其应用[J]．电网技术,2017,41(2):637-643.
• [5]邵冠宇,王慧芳,吴向宏,等.基于依存句法分析的电力设备缺陷文本信息精确辨识方法[J].电力系统自动化,2020,44(12):178-185.
• [6]田园,马文.基于Attention-BiLSTM的电网设备故障文本分类[J].计算机应用,2020,40(增刊2):24-29.
• [7]冯斌,张又文,唐昕,等.基于BiLSTM-Attention神经网络的电力设备缺陷文本挖掘[J].中国电机工程学报,2020,40(增刊):1-10.
• [8]闪鑫,戴则梅,张哲,等.智能电网调度控制系统综合智能告警研究及应用[J].电力系统自动化,2015,39(1):65-72.
• [9]汪崔洋,江全元,唐雅洁,等.基于告警信号文本挖掘的电力调度故障诊断[J].电力自动化设备,2019,39(4):126-132.
• [10]施正钗,郑俊翔,周泰斌,等.基于语义分析的设备监控告警信息知识图谱构建研究[J].浙江电力,2020,39(8):123-129.
• [11]徐家慧,张昊,肖林朋,等.基于GRU神经网络的电网告警信息分类研究[J]．计算机与数字工程,2019,47(6):1405-1408.
• [12]MIKOLOV T,SUTSKEVER I,CHEN K,et al.Distributed representations of words and phrases and their compositionality[C].Proceedings of the 27th Annual Conference on Neural Information Processing Systems.Red Hook:Curran Associates,2013:3113-3119.
• [13]杜修明,秦佳峰,郭诗瑶,等．电力设备典型故障案例的文本挖掘[J]．高电压技术,2018,44(4):1078-1084.
• [14]ZHOU H,FANG Z J,GAO Y B,et al.Feature fusion network based on attention mechanism for 3D semantic segmentation of point clouds[J].Pattern Recognition Letters,2020,133:327-333.
• [15]何晓峰,林子钊,徐希,等.基于模糊化事故等级指标的调度风险评价方法[J].电力系统保护与控制,2021,49(5):98-104.
• [16]王英英,罗毅,涂光瑜,等.基于事故链的电力系统连锁故障风险模糊综合评估方法[J].中国电机工程学报,2010,30(增刊1):25-30.
• [17]王一枫,汤伟,刘路登,等.电网运行风险评估与定级体系的构建及应用[J].电力系统自动化,2015,39(8):141-148.
• [18]LIU C,WANG W,WANG M,et al.An efficient instance selection algorithm to reconstruct training set for support vector machine[J].Knowledge-Based Systems,2017,116:58-73.
• [19]贾志淳,辛民栋,李彦谚,等.基于相似度计算的网络攻击分类方法[J].渤海大学学报(自然科学版),2020,41(2):169-177.