摘要
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 。
Because of stronger background signal and noise signal in measured signal of rub impact fault between rotor and stator in rotor system, it is difficult to obtain precise diagnosis result when wavelet transform is used to diagnose this fault. Aiming at this problem, a new method of processing signal is presented, that is, the track matrix of attractor is firstly reconstructed by time series of the testing signal, and singular value decomposition about this matrix is performed to eliminate background signal and extract remnant signal, then the remnant signal is processed using wavelet transform. The simulated and tested signals are processed and analyzed using this method respectively. The conclusion is that this method is very effective, and can be used to diagnose rub impact fault between rotor and stator in a rotor system precisely.
出处
《振动工程学报》
EI
CSCD
2000年第1期71-77,共7页
Journal of Vibration Engineering
关键词
转子
故障检测
信号处理
奇异值分解
小波变换
rotor
fault detection
signal processing
singular value decomposition
wavelet transform