轨道不平顺历史数据里程偏差修正研究

轨道不平顺历史数据之间存在里程偏差,若直接用于轨道不平顺劣化趋势模型必将导致其预测精度低的效果,从而增加轨道预防性维修的难度。因此,解决轨道平顺状态检测数据之间的里程偏差问题是实现预防性维修的前提。既有研究假设里程偏差为常数,实际上里程偏差随机产生且随着里程增加而累加。研究建立基于钢轨接头匹配的不平顺数据里程偏差修正模型,从轨检仪右高低原始弦数据中使用均一阈值提取出钢轨接头,对齐两段数据上提取到的所有钢轨接头后利用Dynamic Time Warping(DTW)算法,修正检测数据之间的同一相邻两钢轨接头之间的不平顺指标,修正后的检测数据相关系数约为0.98,保证了历史数据运用的可靠性。

Research on Correction of Mileage Deviation of Historical Data of Track Irregularity

There is a mileage deviation between track irregularity historical data. If directly used by the track irregularity degradation trend model, it will lead to low prediction accuracy, which increases the difficulty of preventive maintenance of the track. Therefore, solving the problem of mileage deviation between test data is a prerequisite for implementing preventive maintenance. Existing studies assume that the mileage deviation is constant, and the mileage deviation is, in fact, randomly generated and accumulated as the mileage increases. The research establishes the correction model of the irregularity of the data based on the matching of the rail joints. The uniformity threshold is used to extract the rail joint from the right upper and lower original chord data of the rail detector. After aligning all the rail joints extracted from the two pieces of data, the Dynamic Time Warping(DTW) is used to correct the irregularity index between the adjacent two rail joints between the detected data. The corrected correlation coefficient of the detected data is about 0.98, which ensures the reliability of the historical data.