供水管网中高频压力数据异常模式识别的研究与案例分析  

Research&Case Analysis on Abnormal Pattern Recognition of High-Frequency Pressure Data in Water Supply Network

作　　者：赵云璐 信昆仑[1] Pierre BONARDET 黄慰忠 ZHAO Yun-lu;XIN Kun-lun;Pierre BONARDET;HUANG Wei-zhong(School of Environmental Science&Engineering,Tongji University,Shanghai 200092,China;Centre Technology de Distribution,SUEZ,Le Pecq78230,France;Shanghai Urban Construction Design&Research Institute(Group)Co.,Ltd.,Shanghai 200125,China)

机构地区：[1]同济大学环境科学与工程学院,上海200092 [2]Centre Technique de Distribution,SUEZ,Le Pecq78230,France [3]上海市城市建设设计研究总院(集团)有限公司,上海200125

出　　处：《中国市政工程》2022年第5期108-111,115,129,130,共7页China Municipal Engineering

摘　　要：研究根据实际案例,介绍1种高频压力数据异常模式识别组合模型,组合模型包含异常模式提取、基于DTW距离的相似搜索、基于DTW的K-means时间序列聚类。有望在实际运用中,针对供水管网中已知的异常模式,进行识别预警。选取某压力监测点的高频压力数据对模型进行分析。结果表明,模型能够有效识别出实际案例中的异常模式,归一化互信息(NMI)指数可达0.818。According to the actual case,the research introduces a combination model of abnormal pattern recognition for high-frequency pressure data,which includes abnormal pattern extraction,similarity search based on DTW distance,and K-means time series clustering based on DTW.It is expected to identify&warn the known abnormal patterns in the water supply network in practical application.Select the highfrequency pressure data of a pressure monitoring point to analyze the model.The results show that the model can effectively identify abnormal patterns in actual cases,and the normalized mutual information(NMI)index can reach 0.818.

关 键 词：供水管网 高频压力数据 异常模式识别 时间序列聚类 

分 类 号：TU991.33[建筑科学—市政工程]