基于定子电流监控的轴承故障在线监测

Online approach for bearing fault detection in induction motor using stator current monitoring

中文关键词  轴承故障  故障诊断  时域平均方法  信号分析

英文关键词  bearing fault  fault detection  time domain average  signal analysis

中文摘要

      随着自动化技术的增长, 对感应电动机内部滚动轴承的状态监控得到了飞速发展. 很多优秀的技术手段被运用到轴承故障的在线监测上, 然而还存在着两点不足: 1) 提取的故障信号不够准确; 2) 无法满足在线的需求. 本文在基于定子电流监测的基础上, 提出了一种新型的轴承故障在线诊断方法. 为了能够从电流频谱中提取更加准确的信息, 作者将改进了的时域平均方法(time domain average method, TDA) 运用到故障信号的隔离中. 另一方面, 极限学习机(extreme learning machine, ELM)在分类问题上表现出良好的 泛化性能, 它快速的训练速度能够保证在线故障监测的实施. 文章最后考虑了3种电机运行状态, 仿真结果均证明了此方法的有效性和稳定性.

英文摘要

      Condition monitoring of rolling element bearing faults in induction motors is a fast developing technology during the past decades. Although many excellent technical approaches have been applied to the bearing fault detection, there are two drawbacks: 1) The extracted fault signal is not sufficiently accurate; 2) The approaches are not able to meet the demand of online implementation. This paper proposes a new online bearing fault detection method based on induction motor current monitoring. In order to extract more accurate information from current spectrum, an improved time domain average method is employed to isolate the fault signals. On the other hand, extreme learning machine as a classifier plays an important role in identifying the bearing faults, providing an foundation for online fault detection because of its fast training speed. The simulation results under three operation conditions clearly illustrate the effectiveness and stability of this scheme.