摘要
[目的]为了解决电气化铁路接触网现有检测方式效率低、使用要求高的问题,特对接触网拉出值的检测方式进行研究。[方法]提出了一种将连续小波分解与随机森林相结合的检测方法,旨在实现对不同速度等级下接触网拉出值的有效检测。采用连续小波分解方法对滑板两端的吹响加速度信号进行时频分解,获得时域信号在各个频率区间的分布情况,并在相关频率范围内计算其功率。基于该功率值,构建了一个与拉出值变化趋势相关联的指标。分析了该指标在不同弓头质量、弓头悬挂刚度、高度及运行速度下的变化趋势,发现运行速度对指标具有显著影响。随后,将计算得到的指标及运行速度作为输入特征,训练、测试并验证了随机森林模型。[结果及结论]该方法能准确检测各速度下的接触网拉出值,均方根误差低至5.84,性能优于其他现有算法。研究有效提升了接触网拉出值的测量效率,规避了传统方法对运行路况环境有一定要求的限制,为接触网拉出值检测提供了一种新的思路。
[Objective]To address the inefficiency and high operational requirements of existing detection methods for catenary stagger value in electrified railways,special research on catenary stagger value detection method is conducted.[Method]A detection method combining continuous wavelet transform(CWT)and random forest(RF)is proposed to effectively detect catenary stagger value across different speed levels.CWT is applied to the blowing acceleration signals at both ends of the pantograph contact strip,enabling time-frequency decomposition to reveal the distribution of time-domain signals across various frequency intervals.The power within relevant frequency ranges is calculated,and an indicator correlated with the trend of stagger values is developed based on these power values.The variation of this indicator is analyzed under different conditions in terms of pantograph head mass,pantograph head suspension stiffness,height,and operating speed,revealing that operating speed has a significant impact.Subsequently,the computed indicator and operating speed are used as input features to train,test,and validate an RF model.[Result&Conclusion]The proposed method accurately detects catenary stagger values across various operating speeds,achieving a root mean square error as low as 5.84,outperforming other existing algorithms.In brief,the study effectively enhances the measurement efficiency of catenary stagger values,avoiding the limitations imposed by previous methods on operational environmental conditions,and provides a novel approach for catenary stagger value detection.
作者
陈鸿明
周宁
鲁文伟
杨子贤
程尧
王冬
张卫华
CHEN Hongming;ZHOU Ning;LU Wenwei;YANG Zixian;CHENG Yao;WANG Dong;ZHANG Weihua(State Key Laboratory of Rail Transit Vehicle System,Southwest Jiaotong University,610031,Chengdu,China;CRRC Changchun Railway Vehicles Co.,Ltd.,130062,Changchun,China)
出处
《城市轨道交通研究》
北大核心
2025年第3期236-241,共6页
Urban Mass Transit
基金
国家自然科学基金项目(52072319)
中国国家铁路集团有限公司科技研究开发计划项目(P2020J025,P2021J003,K2021J004-B)。
关键词
轨道交通列车
状态监测
弓网系统
动力学响应
连续小波变换
随机森林
rail transit train
state monitoring
pantograph-catenary system
dynamics response
continuous wavelet transform
random forest
