基于LSTM网络高速列车悬挂系统故障预测方法研究

高速列车悬挂系统安全性对于整车安全运行非常重要,对悬挂系统进行全生命周期状态监测以预测其故障成为不可忽略的研究内容。当高速列车悬挂系统发生机械故障时,产生振动加速度,信号呈现非线性、非平稳特征。文章提出一种基于LSTM网络的高速列车悬挂系统故障预测方法,通过SIMPACK建立某型号高速列车整车模型,获得重要部件在各健康状态下振动信号,以时域指标中均方根值和峰度值指标作为特征构建健康因子曲线,通过LSTM网络对时序数据进行分析预测,为高速列车悬挂系统故障预测提供可行方法,以期为减少悬挂系统故障提供新途径。

On fault prediction method of high-speed train suspension system based on LSTM network

Safety of high-speed vehicle suspension system is very important for the safe operation of the whole train,and the whole life cycle health monitoring of suspension system to predict its failure has become a research detail that must not be ignored.When mechanical failure occurs in the suspension system of high-speed vehicle,the vibration acceleration signal is nonlinear and non-regularity.This paper proposes a fault prediction method of high-speed train suspension system based on LSTM network.It establishes a whole vehicle model of a certain type of high-speed train by using SIMPACK,obtaining the vibration signals of important components in each health state.The health factor curve is constructed by taking the root mean square value and kurtosis value of time domain index as characteristics,and the time series data are analyzed and predicted by LSTM network to provide reference for highspeed train suspension system.This paper provides a feasible method for system fault prediction,in order to provide a new way to reduce the suspension system fault.