具有更新机制的铁路轨道不平顺灰色预测模型  

Grey prediction model for track irregularity with update mechanism

作　　者：郭然[1] 韩宝明[1] 李得伟[1] 李华[1] 

机构地区：[1]北京交通大学交通运输学院,北京100044

出　　处：《中南大学学报（自然科学版）》2013年第10期4334-4341,共8页Journal of Central South University:Science and Technology

基　　金：国家科技支撑计划项目(2009BAG12A10);红果园人才计划项目(I12K00100);教育部基本科研项目(I11JB00020)

摘　　要：为了准确预测轨道各项不平顺的劣化情况,保证轨道的高平顺性,杜绝安全隐患,提出基于灰色预测理论的铁路轨道局部不平顺预测方法。该方法将25 m轨道单元区段各项不平顺幅值的标准差作为原始时间序列,建立具有更新机制的轨道不平顺灰色GM(1,1)非等时距预测模型,并优化模型的背景值和初值以提高预测精度。利用京九线轨检车波形数据,从时间和空间维度验证了模型的有效性。研究结果表明:单元区段预测结果的平均绝对误差为0.039 mm,平均相对误差为1.95%;连续区段预测结果的平均绝对误差为0.046 7 mm,平均相对误差为3.62%。该模型预测不平顺状态与实测结果非常逼近,能够较好地反映轨道局部不平顺的劣化发展。In order to accurately predict the deterioration of track irregularities, ensure high smooth of the track and then fundamentally eliminate hidden dangers, a prediction method for track local irregularities based on grey theory was proposed. With the standard deviation of 25 m track section irregularities＇ amplitudes as the original time series, a grey GM （1, 1） non-equal time interval model with update mechanism was established. To improve the prediction accuracy, the background value and starting value in the model were optimized. With the wave data of track geometry inspection car from Beijing-Kowloon Railway, the prediction validity of the model was verified from time dimension and spatial dimension. The experimental result shows that the mean absolute error of unit section predicting outcomes is 0.039 mm and the mean relative error is 1.95%. For continuous unit section, the statistics are 0.046 7 mm and 3.62% respectively. The prediction irregularities are quite similar to real inspection data, and this model can well reflect the track local irregularities deterioration development.

关 键 词：轨道不平顺 灰色预测理论 GM(1 1) 非等时距 预测模型 

分 类 号：U213.2[交通运输工程—道路与铁道工程]