【作者】 梁瑜；

【导师】 贾利民； 蔡国强；

【作者基本信息】 北京交通大学 ， 系统分析与集成， 2014， 博士

【摘要】 我国地铁的快速发展,大规模建设和投入运营,极大改善我国城市交通现状的同时,也给保障地铁安全带来了挑战。在地铁车辆的故障统计中,机械故障比例最高,而轴承故障在车辆机械故障中占有较高的比重。轴承故障不仅会影响地铁车辆的行车安全,还可能带来人员伤亡和经济损失,造成严重的社会影响,因此非常有必要研究轴承在途故障诊断系统。国内试制的轴承故障诊断系统虽然不少,但是由于系统本身缺陷,最终形成产品并安装到轨道交通车辆上在线监控轴承状态的并不多,安装到地铁列车上的诊断装备则更少,基于以上考虑,本文重点研究适合于地铁列车轴承在途故障诊断的方法与系统,主要完成了以下几方面的研究工作：1.提出了基于自适应傅里叶分解的轴承故障诊断方法。应用自适应傅里叶分解的方法处理轴承原始采集信号,将振动信号自适应分解为一系列单一分量信号,利用峭度值提取包含有故障信息的部分单一分量进行求和,然后对求和的信号做希尔伯特变换,进而做快速傅里叶变换,得到轴承信号的频谱,观察分析频谱图,诊断轴承故障。最后用实验数据验证了方法的有效性；2.提出了基于新的非线性动力学模型的轴承故障诊断方法。根据轴承的物理结构和受力分析,利用弹簧质量阻尼系统首先建立轴承单个滚子与内圈、外圈的接触模型,分析在径向载荷力作用下,内圈、外圈和单个滚子在其法线方向上的实时位移；然后结合随机力干扰技术,建立内圈、外圈和滚子的轴承系统非线性动力学模型,并用该模型计算出在综合力作用下无故障振动的位移和加速度信号。在非线性动力学模型内植入内圈、外圈单表面故障,计算故障情形下的振动位移和加速度信号。最后应用自适应傅里叶分解的方法提取轴承模型内圈、外圈故障仿真加速度信号频谱特征,并与轴承故障特征频率相对比,验证滚动轴承非线性动力学故障模型的有效性；3.提出了基于时频域特征参数融合的轴承故障在途诊断算法。分析了轴承时域特征参数和频域特征参数用于诊断轴承故障的优势和缺点,考虑到时域参数与轴承故障程度密切相关,频域参数可以准确地诊断轴承故障并分辨轴承的故障类型,提出了基于时频域特征融合的轴承故障在途诊断方法,新方法利用神经网络作为非线性分类器识别轴承的在途运行状态,不仅可以诊断轴承是否发生故障,还可以在一定程度上判断轴承的故障程度,同时保留了时域参数和频域参数的优势,最后通过实验数据验证了方法的有效性；4.结合广州地铁列车现有的监控设备以及基于时频域特征参数融合的故障诊断算法,设计了地铁列车轴承故障在途诊断系统,并通过试验台数据验证了系统算法的有效性与实时性。更多还原

【Abstract】 The rapid development, large-scale construction and operation of subway in our country has greatly improved the current situation of our urban traffic, but it also brings challenges to the subway security at the same time. The proportion of mechanical failure is highest in subway vehicle fault statistics, and bearing failure occupies a higher proportion of mechanical failure. Bearing fault not only affects the driving safety of subway vehicle, it may also lead to casualties and economic losses, and causes serious social impact, so is necessary to research on-board diagnosis system of bearing fault.There are many bearing fault diagnosis systems of domestic trial-manufacture, but because of their own defects, very few of them are eventually formig products and installed on the rail transit vehicle to monitor the bearing operative state, so the diagnosis product installed on the subway train is much less. The paper emphasizes on the research of the bearing on-board fault diagnosis technologies and system of urban rail transit, and carries out the following task:1. A new bearing fault diagnosis method is proposed based on Adaptive Fourier decomposition. Adaptive Fourier decomposition is applied to decompose the original bearing vibration signal into a series of mono-components, then part of mono-components with fault imformation is extracted and summed as the bearing fault signal, which is processinged by demodulated resonance technology, and finally the the bearing fault is identified from the obtained frequency spectrum.2. A new bearing fault diagnosis model is proposed. The bearing model is constructed based on the bearing physical structure and force analysis. The contact of each bearing component is supposed as a sping-mass-damping system, and the affect of the number change of the pressure-bearing rollers is considered to contructed the bearing model. Then the bearing fault is described as an extra force and embedded into the bearing model. Finally the bearing fault diagnosis method based on adaptive Fourier decomposition is applied to verify the effectiveness of the proposed model.3. The paper proposes a new bearing fault diagnosis method based on the fusion of time-domain parameters and frequency-domain parameters. It analyzes the advantage and disadvantage of time-domain parameters and frequency-domain parameters applied to diagnose the bearing fault, and proposes a new bearing fault diagnosis method based on the fusion of time-domain parameters and frequency-domain parameters, which has both the advantage of the two kinds of parameters. The new bearing fault diagnosis method is verified by experiment data.4. The paper designs the bearing fault on-road diagnosis system of urban rail transit vehicle by combining the existing monitoring equipment on Guangzhou subway trains and the fault diagnosis algorithm based on the fusion of time-domain parameters and frequency-domain parameters. Finally the effectiveness and real-time of the system’s algorithm is verified by experiment.更多还原