【作者】 杨鑫；

【导师】 姜学东；

【作者基本信息】 北京交通大学 ， 电气工程， 2012， 硕士

【摘要】 轴箱轴承作为列车转向架的关键部件之一,其服役状况对行车安全有着重大影响。实时有效的轴箱轴承在线监测系统不仅可以避免列车事故的发生,还能实现以状态修来替代目前普遍采用的时间修的列车轴箱轴承维修机制,从而降低列车的运营成本。本文以列车转向架轴箱轴承服役过程监测与故障诊断系统为研究对象,依据系统各个子系统的功能特性,开展了轴箱轴承振动信号采集、信号处理分析、故障特征提取以及故障模式识别等方面的深入研究。信号采集实现了列车轴箱轴承振动信号及速度信号的拾取。’论文研究了振动信号的特性,实现了振动信号放大以及基于二阶切比雪夫和二阶巴特沃兹滤波器级联的抗混叠滤波器设计,并根据故障诊断系统信号分析的需求,设计了基于AD7608的模数转换电路。信号处理分析实现了轴箱轴承振动信号消噪和故障模态分析。本文采用了基于小波变换的方法实现振动信号中的消噪处理,有效地去除了噪声对故障信号的干扰。Hilbert-Huang变换是基于信号自身特性的自适应的时频分析方法,本文针对轴箱轴承振动信号非线性、非平稳性的特性,提出了基于Hilbert-Huang变换的列车轴箱轴承故障诊断方法。该方法采用Hilbert-Huang变换对信号进行EMD分解和谱分析,得到了具有明显故障特征的Hilbert谱和Hilbert边际谱,结果表明Hilbert-Huang变换可以有效地对故障进行识别。故障特征提取是提取能有效反映轴箱轴承故障的特征向量。本文对列车轴箱轴承的故障振动信号特征进行了深入研究,提取了能够表征不同故障类型的多特征参量。这些特征参量包括时域统计参数、固有模态函数能量矩以及故障特征幅值比,能够全面地反映轴箱轴承的健康状况。最后论文设计了基于BP神经网络的故障分类器用于故障模式识别。BP神经网络实现了轴箱轴承多特征参量到故障状态之间的非线性映射,对轴承正常、外圈故障、内圈故障及滚动体故障四类工况进行有效识别,实验结果表明,诊断正确率达到了90%以上。更多还原

【Abstract】 Axlebox bearing is one of the key components of train bogie. The condition of the axlebox bearing has great influence on rail traffic security. A real time and effective axlebox bearing fault diagnosis method could avoid traffic accidents and reduce operation costs. By this way, we can adopt condition-based maintenance instead of preventive maintenance which is now being widely adopted. This thesis takes service process monitoring and fault diagnosis system as the research object, studies the signal acquisition, signal processing analysis, fault feature extraction and pattern recognition.Signal acquisition implements the function of packing up vibration and speed signal. Anti-aliasing filter based on second order chebyshev filter and butterworth filter and AD conversion circuit based on AD7608are designed.Signal processing analysis implements the function of vibration signal denoising and fault modal analysis. The thesis adopts wavelet denoising method to remove noise in vibration signal. Hilbert-Huang Transform is an adaptive time-frequency analysis method based on the signal itself. Aiming at the characteristic of vibration signal, an axlebox bearing fault diagnosis method based on Hilbert-Huang Transform is proposed. The Hilbert spectrum and Hilbert marginal spectrum with obvious fault features could be obtained by Hilbert-Huang Transform. The results show that the fault diagnosis method based on Hilbert-Huang Transform can effectively identify the fault.Fault feature extraction implements the function of eigenvectors extraction which could reflect the fault of axlebox bearing. This thesis extracts multi-feature parameters including time domain parameters, energy moments of IMF and fault characteristic amplitude ratio, which could reflect the health status of the axlebox bearing entirely.Fault classifier based on BP Neural Network is designed for fault pattern recognition to realize the nonlinear mapping between multi-feature parameters and fault state. To classify different conditions of axlebox bearing, the diagnostic accuracy is above90%.更多还原