异常值分析在路基连续压实相关性校验中的应用

针对异常点对铁路路基连续压实相关性校验结果的影响问题,在分析异常数据产生原因的基础上,建立数学模型,提出了有效的异常值识别与剔除方法。基于平均统计距离法与邻域估计法,建立二维异常点识别模型,提出了针对相关性校验数据的异常值剔除准则。使用该法对沪昆客运专线3个施工段铁路路基的连续压实检测数据进行分析,结果表明,能够有效识别异常数据,真实反映铁路路基的连续压实检测值,且不被误判。

Application of Outliers Analysis in Correlation Test for Continuous Compaction of Subgrade

In relation to the influence of outliers on the result of correlation test for continuous compaction of railway subgrade, a mathematical model was established by analyzing why the abnormal data arose with an effective outliers identification and elimination method put forward;and depending on average statistical distance method and neighborhood estimation method, a two-dimensional outliers identification model was established and an outliers elimination principle, which is in connection with correlation test data, was proposed.By use of this method, the measurements of continuous compaction of railway subgrade in three construction sections of Shanghai-Kunming Passengers Dedicated Line were analyzed.The result shows that the abnormal data can be recognized effectively and the measurements of continuous compaction of railway subgrade may be reflected truly without misjudgement.