无砟轨道高低不平顺季节性特征及预测

Seasonal Characteristics and Prediction of Longitudinal Irregularity in Ballastless Track

摘要:

为了解高速铁路无砟轨道高低不平顺的季节性特征，基于高速综合检测车实测数据，统计四季条件下CRTSⅠ、Ⅱ、Ⅲ型无砟轨道在不同类型区段的高低不平顺，并利用季节指数法定量评估季节变化对高低不平顺的影响。考虑季节性特征，利用季节指数调整的滑动平均模型，以及深度学习领域的LSTM模型，对CRTSⅢ型无砟轨道的高低不平顺进行预测。结果表明：CRTS系列无砟轨道高低不平顺具有明显的季节性，夏季高低不平顺值普遍高于其他季节。曲线区段高低不平顺受季节变化的影响大于直线区段，桥梁区段高低不平顺受季节变化的影响大于路基区段。季节指数调整的滑动平均模型和LSTM模型均可预测无砟轨道高低不平顺，其中LSTM模型的预测效果更佳。

关键词:无砟轨道;高低不平顺;季节性;季节指数;深度学习;预测模型

Abstract:

In an attempt to investigate the seasonal characteristics of longitudinal irregularity of the ballastless track in high-speed railways， this study utilized actual data collected from the high-speed comprehensive trains to analyze the longitudinal irregularity of China Railway Track System （CRTS） I， II， and III ballastless tracks in different types of sections during the four seasons. The impact of seasonal changes on the longitudinal irregularity was quantitatively evaluated using the seasonal index. Considering the seasonal characteristics， a seasonal index adjusted moving average model， and a Long Short-Term Memory （LSTM） model in the field of deep learning were applied to predict the longitudinal irregularity of CRTS III ballastless track. The results show that the longitudinal irregularity of CRTS series ballastless track has obvious seasonality， with higher values in the summer than in other seasons. The longitudinal irregularity of curved sections is more affected by seasonal changes than that of straight sections， while the longitudinal irregularity of bridge sections is more affected by seasonal changes than that of embankment sections. Both the seasonal index adjusted moving average model and the LSTM model can predict the longitudinal irregularity of ballastless track， with the latter showing better predictive performance.

Key words:ballastless track;longitudinal irregularity;seasonality;seasonal index;deep learning;prediction model