摘要
针对地铁列车塞拉门下挡销、压轮故障,文章提出了一种基于随机森林算法与时域特征分析的检测方法。文章首先对塞拉门在正常及下挡销、压轮故障等不同状态下的电机电流信号进行预处理,并提取预处理后的信号时域特征,建立故障检测的随机森林模型;然后合理设置随机森林模型中的参数,并基于时域特征重要性分析进一步筛选合适的时域特征;最后利用某地铁车辆段现场测试的相应状态下的电机电流数据进行实例分析,对模型的准确性进行验证。结果表明,该方法可实现对地铁列车塞拉门下挡销、压轮故障的有效检测。
A detection method based on random forest algorithm and time domain analysis is proposed for the fault of lower stop pin and pressing wheel of the metro vehicle sliding plug door.Firstly,this paper preprocesses the motor current signals of the sliding plug door under normal condition,the fault of lower stop pin and pressing wheel,and extracts the time domain features of the signals after preprocessing to establish the random forest model for fault detection.Then,the parameters in the random forest model are set reasonably,and the appropriate time domain features are further screened based on the importance analysis of time domain features.Finally,the motor current data under the corresponding states of field test in a car depot are used for example analysis to verify the accuracy of the model.The results show that this method can detect effectively the fault of lower stop pin and pressing wheel of the metro vehicle sliding plug door.
作者
王琳
林川
刘东
WANG Lin;LIN Chuan;LIU Dong(School of Electrical Engineering,Southwest Jiaotong University,Chengdu 611756,China)
出处
《电力机车与城轨车辆》
2022年第5期90-95,共6页
Electric Locomotives & Mass Transit Vehicles
关键词
地铁列车
塞拉门
下挡销
压轮
随机森林算法
故障检测
metro train
sliding plug door
lower stop pin
pressing wheel
random forest algorithm
fault detection
作者简介
王琳,在读硕士研究生,研究方向为轨道交通车辆故障检测。
