| 引用本文: | 汪崔洋,江全元,唐雅洁,朱炳铨,项中明,唐剑.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): |
| WANG Cuiyang,JIANG Quanyuan,TANG Yajie,ZHU Bingquan,XIANG Zhongming,TANG Jian.Fault diagnosis of power dispatching based on alarm signal text mining[J].Electric Power Automation Equipment,2019,39(4): |
|
| 摘要: |
| 电力调度系统在电力系统故障过程中会收到大量告警信号，若调度员无法在短时间内做出决策，则可能使故障扩大，为此提出基于告警信号文本挖掘的电力调度故障诊断方法，该方法包括告警信号文本预处理和故障诊断2个阶段。在第一阶段，基于隐马尔可夫模型(HMM)对告警信号文本进行分词并去除其中的停用词以构建本体词典，并采用向量空间模型(VSM)使文本向量化；在第二阶段，使用滑动时间窗读取实时告警信号，提出一种2层算法，第一层采用支持向量机(SVM)对滑窗内的告警信号进行分类，若分类结果判断为发生故障，则启动第二层k-均值聚类法提取较高可能性的故障供调度员参考。以某电力调度系统实际告警信号作为算例，验证了所提方法的可行性。 |
| 关键词: 电力调度 文本挖掘 向量空间模型 支持向量机 k-均值聚类 |
| DOI：10.16081/j.issn.1006-6047.2019.04.019 |
| 分类号:TM761 |
| 基金项目:国网浙江省电力有限公司科技项目(5211HZ15018Y) |
|
| Fault diagnosis of power dispatching based on alarm signal text mining |
|
WANG Cuiyang1, JIANG Quanyuan1, TANG Yajie1, ZHU Bingquan2, XIANG Zhongming2, TANG Jian3
|
|
1.College of Electrical Engineering, Zhejiang University, Hangzhou 310027, China;2.State Grid Zhejiang Electric Power Company, Hangzhou 310007, China;3.State Grid Hangzhou Electric Power Company, Hangzhou 310009, China
|
| Abstract: |
| The power dispatching system receives massive alarm signals during the failure process of power system, and the failure range may expand if the dispatcher cannot make a decision in a short time, so a fault diagnosis method of power dispatching based on alarm signal text mining is proposed, which includes two stages of alarm signal text preprocessing and fault diagnosis. In the first stage, an ontology dictionary is constructed by segmenting the text of alarm signals based on HMM(Hidden Markov Model) and removing the stop words, and VSM(Vector Space Mo-del) is adopted for text vectorization. In the second stage, the sliding time window is used to read the real-time alarm signals, and a two-layer algorithm is proposed. In the first layer, SVM(Support Vector Machine) is adopted to classify the alarm signals in the sliding window, if the classification result justified to be a fault, the k-means clustering method in the second layer is used to extract faults with higher possibility to dispatcher for reference. A practical alarm signal in a power dispatching system is taken as an example to verify the feasibility of the proposed method. |
| Key words: power dispatching text mining vector space model support vector machine k-means clustering |
