基于时间历程的高速铁路轨道不平顺异常值处理算法

Time-history based processing algorithm for abnormal data from track geometry irregularity in high-speed railway

结合时间历程检测数据与统计方法分析轨道不平顺的变化特征,提出基于轨道不平顺特征阈值与变化率的异常值识别模型,并借助线性预测算法建立异常值修复模型。结合某高铁线路轨道不平顺数据,研究结果表明:轨道不平顺特征阈值范围能很好地刻画轨道不平顺的分布趋势及劣化速率;本文模型能有效提高异常值的识别精度,避免将变化率较大的特殊结构处轨道不平顺识别为异常值;修复后数据具有可靠的几何波形,并能有效减少异常值的不利影响,为准确评估轨道几何状态、高效维护高铁线路提供保障。

Considering inspection data of time history,change characteristic of track irregularity was analyzed.Identification model for abnormal data was proposed based on track irregularity threshold range and change rate.Besides,the repairing abnormal data model was established based on linear prediction.Combining the data from track inspection car in a high-speed railway,results show that track irregularity threshold range can describe the irregularity distribution tendency and degradation rate well.The presented identification model is effective to improve the precision of identifying the abnormal data and avoid misjudging track irregularity with large change rate in specific section.The repairing model can ensure the reliable waveform of inspection data and decrease the adverse effects of abnormal data effectively,which can guarantee the accurate evaluation of track geometry condition and efficient maintenance for high-speed railway.