高速铁路桥上无缝道岔结构力学状态预测方法  

Prediction of Structural Mechanical State of Continuously Welded Turnout on High-speed Railway Bridge

作　　者：马卓然[1] 高亮[1] MA Zhuoran;GAO Liang(School of Civil Engineering,Beijing Jiaotong University,Beijing 100044,China)

机构地区：[1]北京交通大学土木建筑工程学院,北京100044

出　　处：《铁道学报》2022年第2期90-96,共7页Journal of the China Railway Society

基　　金：国家重点研发计划(2016YFB1200402);中国铁路总公司科技研究开发计划(2017G010-A)。

摘　　要：为保证高速铁路桥上无缝道岔的长期安全服役,提出一种数据驱动的轨道结构力学状态预测方法。通过监测数据特征分析,对原始气温数据序列化处理。在此基础上,构建遗传算法优化的Elman神经网络时间序列预测模型,以气温最优序列为输入,预测三项影响结构健康状况的状态监测指标。通过对比不同气温输入形式和不同模型的预测性能,验证方法的优越性。结果表明:所建模型的预测值与监测值较为吻合,预测平均绝对误差相较其他模型降幅达39.48%,均方根误差降幅达29.10%,误差标准偏差降幅达10.92%;该模型的预测误差小,预测结果稳定可靠,能够为轨道结构力学状态的准确预测和安全隐患的及时预警提供支持。A data-driven method for the prediction of mechanical state of track structure was proposed,to ensure the long-term service of continuously welded turnout track system on high-speed railway bridge.Through the analysis of the characteristics of monitoring data,the original atmospheric temperature data was serialized.On this basis,a genetic algorithm-optimized time series forecasting model combined with Elman neural network was established,which uses the optimal temperature series as input to predict three monitoring indicators reflecting structural health condition.The superiority of the method was verified by comparing the prediction performance of different temperature input forms and different models.The results show that the predicted values of the proposed model are in good agreement with the observed values,with 39.48%decrease of the average absolute error of prediction,29.10%decrease of the root mean square error,and 10.92%decrease of the error standard deviation,compared with other models.The model shows small prediction error with stable and reliable results,which can provide support for the accurate prediction of the mechanical state of the track structure and timely warning of safety risks.

关 键 词：高速铁路 无缝道岔 结构状态预测 神经网络 遗传算法 

分 类 号：U216.9[交通运输工程—道路与铁道工程]