地铁车辆电动塞拉门的剩余寿命预测  

Prediction of Remaining Useful Life for Electrical Sliding Plug Door of Metro Vehicles

作　　者：王凌[1] 陈长骏[1] 潘静[1] 许宏[1] 陈锡爱[1] 那文波[1] Wang Ling Chen Changjun Pan Jing Xu Hong Chen Xiai Na Wenbo(China Jiliang University, Hangzhou,310018)

机构地区：[1]中国计量大学,杭州310018

出　　处：《中国机械工程》2016年第22期2994-3000,共7页China Mechanical Engineering

基　　金：国家自然科学基金资助项目(51504228);浙江省自然科学基金资助项目(LY14F030019;LQ14F050003)

摘　　要：针对地铁车辆客室电动塞拉门传动装置润滑不良的问题,提出了基于自组织映射(SOM)神经网络、隐马尔可夫链(HMC)模型和蒙特卡罗(MC)仿真的剩余使用寿命预测方法。该方法首先对采集到的电机电流信号进行特征提取;然后利用SOM对提取出的多维特征数据进行融合与编码,将所得结果作为HMC的输入向量,训练得到全部寿命下劣化状态转移矩阵;最后利用MC方法实现对其劣化过程的剩余使用寿命预测。故障模拟实验结果表明,该方法可以在考虑润滑不良故障模式下,有效预测得到电动塞拉门丝杆的剩余使用寿命。To solve the problems of poor lubrication associated with elecrical sliding plug doors of metro vehicles,aprediction model of the remaining useful life was proposed herein based on self-organizing feature map（SOM）,hidden Markov chain（HMC）and Monte Carlo（MC）simulation.Firstly,the motor current signals were collected and the features were extracted.Secondly,the SOM method was used to achieve data fusion and encoding for the multi-dimensional feature data.Then the results were used as input vector of the HMC in order to obtain state transition probabilities for the whole life.Finally,the MC simulation was used to predict the remaining useful life of the degradation process.The fault simulation experimental results show that the method can predict the remaining useful life effectively of the electrical sliding plug door with the consideration of the failure mode of poor lubrication.

关 键 词：电动塞拉门 润滑不良 剩余使用寿命预测 隐马尔可夫链模型 自组织映射神经网络 蒙特卡罗仿真 

分 类 号：TH17[机械工程—机械制造及自动化] TP18[自动化与计算机技术—控制理论与控制工程]