侯迎团,杨卫平,章一洲.基于Chauvenet准则的轨距异常值在线处理[J].测控技术,2016,35(6):56-59 |
基于Chauvenet准则的轨距异常值在线处理 |
On-Line Gauge Outlier Processing Based on Chauvenet Criteria |
|
DOI: |
中文关键词: 轨道检测 轨距 轨缝异常值 Chauvenet准则 Lilliefors检验 |
英文关键词:track inspection gauge rail joints outliers Chauvenet criteria Lilliefors test |
基金项目:中航工业集团创新基金资助项目(2010F61825) |
|
摘要点击次数: 889 |
全文下载次数: 324 |
中文摘要: |
轨道几何参数(如轨距、水平、高低和轨向等)检测是轨道质量状态检测监测的重要内容。然而,检测数据受轨缝等轨道不连续部分的影响,经常会出现检测异常值,从而引起频繁的超限误报。针对轨道检测中轨距异常值问题,分析了实测轨距数据的特点,提出了轨距变化率服从正态总体分布的假设,通过Lilliefors正态性检验验证了该假设的合理性,并提出了基于Chauvenet准则的轨距检测异常值在线处理算法。该算法是一种通用算法,适用于样本自身或预处理(如变化率)后满足正态性假设的情况。 |
英文摘要: |
The inspection of track geometry parameters(such as gauge,level,longitudinal level and alignment) is an important part of the track quality state inspection monitoring.However,test data are affected by discontinuous portion such as rail joints,inspecting outliers often occur to cause frequent overrun false positives.In order to solve the problem of track gauge outlier inspection,the characteristics of the measured gauge data are analyzed,the rate of gauge change obeying normal population distribution assume is proposed,whose reasonableness is verified through the Lilliefors normality test,online gauge outlier processing based on Chauvenet criteria is proposed as well.The algorithm is a general algorithm for the sample itself or after pretreatment(such as rate of change) to meet the assumption of normality. |
查看全文 查看/发表评论 下载PDF阅读器 |
关闭 |