中文摘要：

为排除转子系统故障诊断中各种噪声干扰，提出基于局域波分解尺度滤波来提纯轴心轨迹的方法，并给出尺度滤波的滤波器表达式．局域波法将任何复杂信号分解成为有限数目的具有一定物理意义的基本模式分量（IMFs），每一个分量描述时变信号中不同频率和尺度范围的固有振动模式．通过对加噪信号的有效去噪，检验了尺度滤波的可行性．诊断实例说明，该方法可合理、有效地提取转子系统的故障特征．

英文摘要：

To prevent the interference of noise from blurring the rotor system with fault diagnosis, a purified method based on the time-scale characteristic filtering of local wave was developed, and the formula of filter was introduced. Local wave method can decompose any complicated signal into a finite number of intrinsic mode functions（IMFs） with local wave decompo- sition and defining instantaneous frequency. Every IMF describes different frequency and scale of the inherence oscillation mode. The analysis result shows that the feasibility is proved by filtering of signals with noise and the method is introduced to the identification of axis orbit of misalignment rotor, which could rationally and effectively purify axis orbit and extract the fault signature.