高速列车非平稳振动信号盲源分离方法及应用

发布时间：2018-03-16 21:11

本文选题：盲源分离　切入点：全局最优信噪比算法　出处：《机械工程学报》2014年19期 　论文类型：期刊论文

【摘要】：高速列车具有若干时变激励源,传统的时频分析方法只能对观测的混合振动的总体强度分布、时频域结构加以分析,不能分离出与各振源对应的信号分量从而明晰振源状态与故障特征。盲源分离是一种可行的分析方法,但由于高速列车振动信号具有时变振源数目、时变信号长度、受车速调制的变频非平稳等特征,传统的盲源分离方法不适用。为了提高高速列车非平稳信号的盲源分离效果,基于自适应滤波理论提出全局最优信噪比盲源分离新方法,并对其可分离性的判别依据进行论证。新方法的有效性经仿真计算和实测数据分析得到验证。研究表明:新方法对高速列车时变非平稳信号的盲源分离效果优于传统的基于非线性函数的盲源分离方法和基于高阶累积量的盲源分离方法。
[Abstract]:The high speed train has several time-varying excitation sources. The traditional time-frequency analysis method can only analyze the overall intensity distribution and time-frequency domain structure of the observed mixed vibration. The signal components corresponding to each vibration source can not be separated so as to clarify the state of vibration source and fault characteristics. Blind source separation is a feasible analysis method, but because of the number of time-varying vibration sources and the length of time-varying signal, the vibration signal of high-speed train is time-varying. In order to improve the blind source separation effect of non-stationary signals of high-speed trains, a new global optimal signal-to-noise ratio blind source separation method is proposed based on adaptive filtering theory. The validity of the new method is verified by simulation and analysis of measured data. The results show that the new method is superior to the blind source separation method for time-varying non-stationary signals of high-speed trains. The traditional blind source separation method based on nonlinear function and the blind source separation method based on higher order cumulant.
【作者单位】：西南交通大学机械工程学院;
【基金】：国家自然科学基金(51205323,51275426) 国家科技支撑计划(2009BAG12A01) 中央高校基本科研业务费专项资金(2682014CX032)资助项目
【分类号】：TN911.7

【参考文献】