高速列车牵引电机轴承故障机理及智能诊断方法研究

Bearing is the crucial equipment of traction motor of China Railways High-speed(CRH) train, which is subjected to the joint effects of much more loads, there exist some natural particularitis of the freedom movements (traversing, ups and downs, stretching and rotating)and complex random vibration to easily damage and cause accidents of the broken machine and outage and so on.Traction motor bearing of CRH is selected as research object in this project. The fault dynamic evolution mechanism, feature extraction,intelligence diagnosis and experiment are researched by combinating methods of theoretical analysis,numerical computation and physical tests under the special service environments. By analyzing multiple factors of fault simulation,vibration,noise,temperature and frequency,bearing fault modeling method based on dynamics simulation model and transferring characteristics of fault signals are studied, and the fault dynamic evolution mechanism is clarified and mapping relations between bearing parameters and fault features are indicated. From the point of view of suppressing and utilizing noise,these methods of biomimetic pattern recognition,adaptive stochastic resonance theory and mathematical morphology transform are introduced to preprocess, strengthen and extract features of multi-source weak fault signals under complex route.The dimension reduction reconstruction of feature matrix is also studied in here. Hybrid intelligent diagnosis mtethod with multi-layered fusion is propoded by optimizing hierarchical structure and an adaptive trend prediction mothod based on fault features and data-driven is also propoded in this research. The goal is to provide the theoretical basis and method support for autonomously monitoring and forecasting the health status of traction motor of CRH train.

轴承作为高速列车牵引电机的关键部件，受到承重、传递、冲击等载荷联合影响，存在"横移、沉浮、伸缩、旋转"等自由度运动和复杂的随机振动固有特殊性，使其极易发生损坏，导致机破、停运等事故多发。本项目围绕高速列车牵引电机轴承系统，采用理论分析、数值计算和实验相结合，开展特殊服役环境下轴承故障动态演化机理、特征提取、智能诊断与实验研究。通过分析故障模拟与振动、噪声、温度、转频等多因素影响，研究基于动力学仿真模型的故障建模方法，探索故障信号传递规律，阐明故障动态演化机理，揭示轴承参数与故障状态之间的映射关系；从抑制和利用噪声出发，引入仿生模式识别、自适应随机共振理论和数学形态变换等方法，研究复杂路径下多源微弱故障信号的预处理、增强与特征提取方法，实现特征矩阵降维，提出多层融合的智能诊断方法和故障特征数据驱动的自适应趋势预测方法，为高速列车牵引电机健康监测与预警提供理论基础与方法支撑。

牵引电机作为极端服役环境下高速列车驱动装置的“心脏”，其轴承是最为关键的部件之一，它的运行状态直接决定着高速列车能否安全、高效、可靠运行。因此本项目以高速列车牵引电机为研究对象，引入了以机械为主要载体的多学科知识交叉融合的系统性方法，采用了理论分析、数值计算、实验验证与应用相结合，分析电机轴承故障模拟与振动、噪声、温度、转频等多因素影响，研究了电机轴承系统的非线性动力学建模与仿真方法，建立了电机轴承系统的非线性动力学模型与故障信号的传递模型，阐述了电机轴承系统故障动态演化机理，探索了复杂传递路径下振动信号的传递规律；从抑制和利用噪声出发，研究并提出了基于仿生模式识别的多源微弱故障信号预处理方法、基于改进自适应随机共振理论的多微弱故障信号增强方法、基于EEMD、最优模态和多尺度模糊熵的特征提取及其矩阵构建方法、基于EEMD、相关系数法和希尔伯特变换的特征提取方法和基于信息熵和相关性理论的特征参数敏感性与冗余度评估方法；进而研究并提出了基于故障动态演化机理和信息融合的智能诊断模型与方法、基于改进粒子群和SVM的电机轴承故障诊断方法、基于改进VMD和Hilbert变换的电机轴承故障诊断方法；在此基础上，定义了一种高阶差分数学形态梯度谱熵，研究并提出了基于高阶差分数学形态梯度谱熵的滚动轴承损伤程度识别方法和基于高阶差分数学形态谱熵和极限学习机的性能趋势预测方法，用于评估电机轴承的健康状态。实际数据被用于验证了模型与方法的有效性。本项目的研究成果为高速列车牵引电机的健康监测、智能诊断与预警提供了理论基础与方法支撑。. 受本课题资助，课题组经过四年的深入研究，项目执行情况良好，进展顺利，取得了较好的研究成果。在Journal of Process Mechanical Engineering、IEEE Access、Soft Computing、Entropy等著名期刊发表学术论文43篇，其中SCI期刊论文21篇，EI期刊论文15篇；授权发明专利3项，申请发明专利2项，获得辽宁省自然科学学术成果奖二等奖2篇、三等奖2篇；培养硕士生22人，培养博士生5人。