长大隧道施工中有轨运输轨道高低不平顺估计  

Estimation of track longitudinal level irregularity in the construction of long and large tunnel rail transportation

作　　者：崔泽毅 王建西[1,2,3] 王晓曼 郭庆 邢文佳 CUI Zeyi;WANG Jianxi;WANG Xiaoman;GUO Qing;XING Wenjia(State Key Laboratory of Mechanical Behavior and System Safety of Traffic Engineering Structures,Shijiazhuang Tiedao University,Shijiazhuang 050043,China;Key Laboratory of Roads and Railway Engineering Safety Control of Ministry of Education,Shijiazhuang Tiedao University,Shijiazhuang 050043,China;Hebei Province Railway Coupler System Technology Innovation Center,Shijiazhuang 050043,China)

机构地区：[1]石家庄铁道大学,省部共建交通工程结构力学行为与系统安全国家重点实验室,石家庄050043 [2]石家庄铁道大学,道路与铁道工程安全保障省部共建教育部重点实验室,石家庄050043 [3]河北省铁路扣件系统技术创新中心,石家庄050043

出　　处：《交通科技与经济》2023年第5期60-65,共6页Technology & Economy in Areas of Communications

基　　金：国家重点研发计划项目(2021YFB2601000);河北省自然科学基金项目(E2020210092)。

摘　　要：针对长大隧道施工中有轨快速运输轨道高低不平顺检测方法缺乏便捷高效性问题,通过轴箱垂向加速度和轨道高低不平顺之间的关联关系,并结合长短记忆神经网络(LSTM)和门控循环单元神经网络(GRU),建立LSTM与GRU融合的轨道高低不平顺估计模型。结果表明:利用LSTM-GRU模型估计的高低不平顺幅值和对比值变化趋势基本相同;估计值和对比值的差值最大为1.38 mm,占最大幅值的8.84%;其中差值在15%(1.02 mm)以内的数量占样本总数的86.4%。LSTM-GRU模型相对LSTM模型,训练时间下降37.51%;相对于GRU模型,高低不平顺均方根误差下降30.77%。针对长大隧道施工过程中的有轨运输,LSTM-GRU模型不仅能保证估计精度,还能明显降低估计时间,对长大隧道施工有指导意义。In response to the lack of convenience and efficiency in detecting track longitudinal level irregularity during rapid rail transportation in the construction of long and large tunnel,a track longitudinal level irregularity estimation model integrating LSTM and GRU is established by analyzing the correlation between vertical acceleration of axle box and track longitudinal level irregularity,and combining Long Short Memory Neural Network(LSTM)and Gated Recurrent Unit Neural Network(GRU).The result shows that the variation trend of longitudinal level irregularity contrast value and estimated value by LSTM-GRU estimation model is basically the same;the maximum difference between the estimated value and the contrast value is 1.38 mm,accounting for 8.84%of the maximum amplitude;the number of samples with difference within 15%(1.02 mm)accounts for 86.4%of the total number of samples.Compared with LSTM estimation model,the training time of LSTM-GRU estimation model decreased by 37.51%;compared with GRU estimation model,the root mean square error decreases by 30.77%.The LSTM-GRU model not only ensures estimation accuracy but also significantly reduces estimation time for rail transportation during the construction of long and large tunnels,which has guiding significance for the construction of long and large tunnels.

关 键 词：铁路线路工程 高低不平顺估计 LSTM-GRU模型 长大隧道 有轨快速运输 

分 类 号：U216.3[交通运输工程—道路与铁道工程]