基于神经网络模型的高铁轮对故障诊断和预测方法的研究  

Research on Fault Diagnosis and Fault Prediction Method of High-speed Railway Wheelset Based on Neural Network

作　　者：季银银 刘婷婷 沈建洲 缪响 张颖[1] JI Yinyin;LIU Tingting;SHEN Jianzhou;MIAO Xiang;ZHANG Ying(School of Information and Communication Engineering,Nanjing Institute of Engineering,Nanjing 211167,China;Industrial Center of Nanjing Institute of Engineering,Nanjing 211167,China)

机构地区：[1]南京工程学院信息与通信工程学院,南京211167 [2]南京工程学院工业中心,南京211167

出　　处：《机电工程技术》2020年第5期14-17,共4页Mechanical & Electrical Engineering Technology

基　　金：江苏省高等学校自然科学研究重大项目(编号:19KJA410001)。

摘　　要：提出将高铁轮对运转时产生的振动信号作为样本,分析在高铁轮对运转中,其振动信号中均值、方差、均方值、峰度、裕度因子、脉冲因子等值的变化。由于时域振动信号分析具有很强的实时性,因此采用振动时域信号作为特征信号,提取出能量参数、峰度参数、波形参数、裕度参数、脉冲参数和峰值参数作为样本输入到神经网络模型中,提出利用概率神经网络模型进行高铁轮对故障诊断。利用径向基网络模型,分析历史故障数据,对故障初期显示出的信号特征进行分类,确定中心节点,预测出故障类型,保障高铁轮对可靠运行。It was proposed to take the vibration signal generated during the operation of high-speed railway wheel set as a sample to analyze the changes of mean value, variance, mean square value, kurtosis, margin factor and pulse factor in the vibration signal during the operation of high-speed railway wheel set. Because the time-domain vibration signal analysis is more real-time, the vibration time-domain signal was used as the feature signal to extract the energy parameters,kurtosis parameters, waveform parameters, margin parameters, pulse parameters and peak parameters as samples and input them into the neural network model,and a probabilistic neural network model was proposed for fault diagnosis of high-speed railway wheel sets. Then using the radial basis network model, the historical fault data was analyzed, the signal characteristics displayed in the early stage of the fault were classified, the center node was determined, the fault type was predicted, and the reliable operation of the high-speed rail wheel set was ensured.

关 键 词：神经网络 故障诊断 故障预测 

分 类 号：U269[机械工程—车辆工程]