基于承前序列的坝体位移异常值在线检测方法  

Online Outlier Detection Method of Dam Displacement Based on Previous Sequence

作　　者：花胜强 郑慧娟 陈意 高磊 郑健兵 HUA Shengqiang;ZHENG Huijuan;CHEN Yi;GAO Lei;ZHENG Jianbing(NARI Group Corporation(State Grid Electric Power Research Institute),Nanjing 211106,Jiangsu,China)

机构地区：[1]南瑞集团(国网电力科学研究院)有限公司,江苏南京211106

出　　处：《水力发电》2021年第12期102-105,共4页Water Power

摘　　要：为确保大坝安全监测日常工作的正常开展,针对大坝安全监测中坝体位移异常值的在线判别问题,提出了一种基于承前序列的检测方法。通过基于中位数、绝对中位数偏差的测值分布检验和基于Q值法的波动率检验,平衡了异常值误判和漏判的风险,提高了检测正确率的同时也增强了方法的鲁棒性。实际算例表明,相对于四分位法、格拉布斯准则,新方法精度高、负载小,适用于作为坝体位移异常值在线检测的方法。In order to ensure the normal implementation of routine dam safety monitoring,a method based on the previous data sequence is proposed for the online outlier detection of dam displacement data.By the distribution test based on the median and absolute median deviation and the volatility test based on the Q-value method,the risk of misjudgment and miss-judgment of outliers is balanced with the improved detection accuracy and robustness.The simulation example shows that,compared with the quartile method and Grubbs criterion,this method has high accuracy and low cost,and it is suitable for online outlier detection of dam displacement monitoring.

关 键 词：大坝监测 在线检测 承前序列 Q检验 中位数 

分 类 号：TV698.1[水利工程—水利水电工程]