由于在转子系统动静件碰摩故障测试信号中背景信号、噪声信号较强 ,使得在利用小波变换进行故障诊断时 ,难以得到准确的结果。本文提出了一种信号处理方法 ,即首先利用测试信号的时间序列重构吸引子的轨迹矩阵 ,并且对该矩阵进行奇异值分解以剔除背景信号、提取残余信号 ,然后再对残余信号进行小波变换。利用该方法分别对仿真信号、实测信号进行了处理和分析 ,结果表明本方法是非常有效的 ,可对转子系统动静件碰摩故障进行准确地诊断 Due to the background signal and the strong noise signal in the rubbing fault test signal of the rotor system, it is difficult to obtain accurate results when using the wavelet transform to diagnose the fault. In this paper, a signal processing method is proposed. Firstly, the track matrix of the attractor is reconstructed by using the time series of the test signal, and the matrix is ​​singularly-valued decomposed to reject the background signal, the residual signal is extracted, and then the residual signal is subjected to wavelet transform . The method is used to process and analyze the simulated signal and the measured signal respectively. The results show that this method is very effective and can accurately diagnose the rub-impact fault of the rotor system