基于时频分析的欠定信号盲分离与微弱特征提取

发布时间：2018-01-06 10:27

本文关键词：基于时频分析的欠定信号盲分离与微弱特征提取　出处：《机械工程学报》2014年18期 　论文类型：期刊论文

【摘要】：盲源分离对于多振源信号的故障诊断与识别是一种有效的方法,但是传统的盲源分离算法都是针对观察信号大于或等于源信号的情况,但对于观察信号小于源信号的欠定盲分离问题,这在很大程度上制约了盲源分离的实际应用。通过应用经验模式分解和时频分析对非平稳信号分析的优势,提出基于时频分析的欠定盲源分离方法进行设备微弱特征提取。对振动信号进行经验模式分解,并根据分解得到的内蕴模式分量估计源信号个数并选择最优的观察信号,将振动信号与选择的最优观察信号组成新的观察信号进行基于时频分析的盲源分离,通过对仿真信号和齿轮箱实测信号进行验证分析。并与基于独立分量分析的盲源分离算法进行对比,研究表明基于时频分析的盲源分离对混合信号具有更好的分离效果,能够较好地对微弱特征进行提取。
[Abstract]:Blind source separation is an effective method for fault diagnosis and identification of multi-source signals, but the traditional blind source separation algorithms are aimed at the situation that the observed signal is greater than or equal to the source signal. However, the application of blind source separation is restricted to a great extent for the problem of under-determined blind separation of observed signal smaller than source signal. The advantage of empirical mode decomposition and time-frequency analysis is applied to non-stationary signal analysis. An under-determined blind source separation method based on time-frequency analysis is proposed to extract the weak features of the equipment and to decompose the vibration signal by empirical mode decomposition. The number of source signal is estimated according to the intrinsic mode component and the optimal observation signal is selected. The vibration signal and the selected optimal observation signal constitute a new observation signal to separate the blind source based on time-frequency analysis. The simulation signals and gearbox measured signals are verified and analyzed, and compared with the blind source separation algorithm based on independent component analysis (ICA). The results show that blind source separation based on time-frequency analysis has better separation effect on mixed signals and can extract weak features better.
【作者单位】：大连理工大学机械工程学院;
【基金】：国家自然科学基金(51175057) 中央高校基本科研业务费专项资金(DUT14ZD204)资助项目
【分类号】：TN911.7
【正文快照】： 0前言*盲源分离可以将振动信号中的众多振动源信号分离出来,从而对机械设备进行故障诊断,是在20世纪80年代中后期发展起来的一种功能强大的信号处理方法,为机械设备多故障复合诊断提供了一种新的方法,已成为很多领域研究和发展的重要课题,同时也具有十分重要的实用价值。测试

【参考文献】