基于改进非等时距灰色组合模型的轨道质量预测方法研究  

Research on Track Quality Prediction Method Based on Improved Unequal Interval Grey Combination Model

作　　者：赵玉林 饶志强 常惠[1] 李益晨 ZHAO Yulin;RAO Zhiqiang;CHANG Hui;LI Yichen(Beijing Key Laboratory of Information Service Engineering,Beijing Union University,Beijing 100101,China;Urban Rail Transit and Logistics College,Beijing Union University,Beijing 100101,China)

机构地区：[1]北京联合大学北京市信息服务工程重点实验室,北京100101 [2]北京联合大学城市轨道交通与物流学院,北京100101

出　　处：《铁道建筑》2020年第11期111-114,127,共5页Railway Engineering

基　　金：国家自然科学基金(61871039);北京联合大学科研项目(WZ20201905,ZK30202001);北京联合大学教育科学研究课题(JK202011);国家级轨道交通通信与控制虚拟仿真实验室开放课题(2019RTCC04);北京联合大学研究生科研创新项目(YZ2020K001)。

摘　　要：轨道质量指数(TQI)是制定轨道维修计划的基础,具有随时间缓慢变化并伴有随机波动的特点,现有模型难以有效预测这种变化。为了提升TQI随时间和条件变化的预测精度,提出一种改进非等时距灰色组合IGM⁃PSO⁃Elman预测模型。通过优化一阶累加过程和背景值计算,在参数求解中植入权重矩阵对传统灰色模型进行了改进,提高了轨道状态的预测精度;利用经粒子群算法优化的Elman神经网络对该模型的预测残差值进行了校正,进一步降低了预测误差。选取平均相对误差、均方根误差、决定系数、相关系数4个评价指标,利用实测数据与文献数据对该组合模型进行了验证,并研究了样本数对预测精度的影响。结果表明:增加样本数可以提高预测精度;该组合模型比传统灰色模型预测精度更高,能够为制定线路状态维修计划提供技术支持。Track quality index(TQI)is the basis of track maintenance planning.It has the characteristics of slow change with time and random fluctuation,and it is difficult for existing models to predict this change effectively.In order to improve the prediction accuracy of TQI changing with time and condition,an improved unequal interval grey combination IGM-PSO-Elman prediction model was proposed.The traditional grey model was improved by optimizing the first-order accumulation process and background value calculation,and embedding weight matrix in parameter solution,which has improved the prediction accuracy of track state.The Elman neural network optimized by particle swarm optimization is used to correct the prediction residual value of the model,which further reduces the prediction error.Four evaluation indexes including average relative error,root mean square error,determination coefficient and correlation coefficient were selected to verify the combination model by using measured data and literature data,and the influence of sample number on prediction accuracy was studied.The results show that increasing the number of samples can improve the prediction accuracy,the combination model has higher prediction accuracy than the traditional grey model,and can provide technical support for the formulation of track maintenance plan.

关 键 词：轨道质量指数 非等时距 IGM⁃PSO⁃Elman 预测残差 模型验证 状态维修 

分 类 号：U216.4[交通运输工程—道路与铁道工程]