轨道不平顺检测数据的预处理方法分析

Analysis of preprocessing methods of track irregularity inspection data

针对数据异常值、低频趋势项以及里程漂移三方面偏差展开分析,采用拉依达准则、轨道不平顺变化率法、EMD高通滤波法和灰色关联度法对检测数据偏差进行处理。结果表明:去除异常值时应首选轨道不平顺变化率法;采用EMD高通滤波的方法可以有效地去除轨道不平顺检测数据中低频趋势项;灰色关联度的方法可以较好地解决多次检查数据里程漂移的问题。最后,采用MATLAB的GUI开发轨道不平顺预处理模块软件,实例分析证明了该软件的有效性。建议推广使用这些数据预处理方法,以提高我国轨道状态养护维修效率和管理技术水平。

Three kinds of deviation,outliers,the low frequency trend term and mileage drift,were analyzed in this paper. PauTa rule,changing rate method of track irregularity,empirical mode decomposition approach and gray correlation analysis method were applied to process deviation of measured data. The results show that track irregularity changing rate method is the best way to remove outliers,and empirical mode decomposition approach is the best way to eliminate the low frequency trend term,while gray correlation analysis way is the best means to resolve mileage drift. At last,preprocessing software of track irregularity measured data was exploited by using GUI technology of MATLAB. The example shows the validity of the developed software. It is recommended that these data preprocessing methods should be popularized in order to improve the efficiency and management level of maintenance and repair of the rail track.