• CN:11-2187/TH
  • ISSN:0577-6686

机械工程学报 ›› 2018, Vol. 54 ›› Issue (7): 54-62.doi: 10.3901/JME.2018.07.054

• 机械动力学 • 上一篇    下一篇

基于听觉显著图的瞬态振动信号表征方法研究

李允公, 戴丽, 李国萌, 王波, 吴文寿   

  1. 东北大学机械工程与自动化学院 沈阳 110189
  • 收稿日期:2017-04-06 修回日期:2017-09-13 出版日期:2018-04-05 发布日期:2018-04-05
  • 通讯作者: 李允公(通信作者),男,1976年出生,博士,副教授。主要研究方向为机械故障诊断。E-mail:ygli@me.neu.edu.cn
  • 作者简介:戴丽,女,1979年出生,博士,讲师。主要研究方向为机电系统动力学。E-mail:ldai@me.neu.edu.cn
  • 基金资助:
    国家自然科学基金资助项目(51275080)。

A Calculation Method of Auditory Saliency Map for Representation of Transient Vibration Signal

LI Yungong, DAI Li, LI Guomeng, WANG Bo, WU Wenshou   

  1. School of Mechanical Engineering & Automation, Northeastern University, Shenyang 110189
  • Received:2017-04-06 Revised:2017-09-13 Online:2018-04-05 Published:2018-04-05

摘要: 瞬态振动信号会使信号的幅值与频率结构发生"显著"变化,也是设备故障的重要标志。将听觉显著图方法引入振动信号分析中,并基于瞬态信号的时频分布特征,提出了一种新的听觉显著图计算方法,以用于表征瞬态信号。首先对信号进行带通滤波、包络提取和非线性压缩,继而模拟听神经元的时频感受野特性,进行多尺度二维和一维低通滤波,并对滤波结果进行跨尺度整合,以凸显发生显著变化的时频区域,最后,综合两种滤波整合结果,得到时频显著图和全局显著图。其中,设计了新的二维低通滤波器和跨尺度整合方法。数值仿真和试验验证结果表明,所提方法能够有效表征瞬态信号占据的时频区域,与已有方法相比,在抗噪声干扰、屏蔽稳态背景信号和微弱瞬态信号表征等方面均具有更为良好的性能。

关键词: 带通滤波, 低通滤波, 故障诊断, 瞬态信号, 听觉模型

Abstract: The transient vibration signal always is a significant symbol of mechanical fault, and can lead to a prominent change on amplitude and frequency structure of signal. A method named auditory saliency map is introduced to vibration signal analysis, and according to the shape characteristic in time-frequency domain of transient signal, a novel calculation method of auditory saliency map is proposed based on the operating mechanism of auditory system. The signal is band-pass filtered by Gammatone filters firstly, and the envelope filter signal is extracted and compressed nonlinearly. Then the multi-scale two-dimension and one-dimension low-pass filter is carried out respectively. To highlight the time-frequency regions, in which the signal change observably, the filter results are integrated between different filter scale. At last, by synthesizing integrate results in two kinds of filter, the time-frequency saliency and global saliency are received. The results of numerical simulation and experimental verification show that the proposed method can effectively represent the time-frequency regions of transient signal, moreover, compared with existing methods, in the aspects of anti-noise, masking stationary background signal and representing weak transient signal, the propose method has better performance.

Key words: auditory model, band-pass filter, faults diagnosis, low-pass filter, transient signal

中图分类号: