改进的随机共振和EMD混合模型用于转子早期故障检测

Conjoint model combining improved stochastic resonance and empirical mode decomposition for rotor incipient faults detection

针对旋转机械转子早期故障检测精确度低的问题,建立了尺度变换随机共振降噪下的经验模式分解(EMD)模型。利用尺度变换随机共振模型在全频段范围内自适应地提取待测信号中所含频率信息,为了避免漏警检测造成的安全隐患,模型选用低阈值检测共振频率,但在强噪声扰动下有可能带来虚警检测;为了去除虚警检测,该模型根据检测到的共振频率对实测信号进行带通滤波,将滤波后的信号通入改进的EMD系统,以检测出共振频率分量对应的幅度值,剔除幅度值较小的虚警现象,从而保证整套模型具有高精确度故障检测的性能。理论分析和实测结果表明,该混合模型能准确检测出旋转机械转子的早期故障信息。与现有方法相比,该混合模型故障检测结果具有更高的可靠性。

In order to achieve the precise detection of the rotor incipient faults of rotating machinery, a conjoint model combining the scaling transform stochastic resonating model and the empirical mode de-composition （EMD） model is presented. It took fully advantage of scale transform stochastic resonance model in adaptively detecting the original noisy signal all-band frequency with low threshold at first. But false-alarm phenomenon may occur in very complicated noise environment. The original noisy signal was processed by a band-pass filter whose passband was designed by the resonance frequencys. The de-noised signal amplitude value was detected by improved EMD model. When resonance signal amplitude values were small, the false-alarm phenomenon happened. Theoretical analysis and experimental results show that the system exactly detects the rotor incipient faults of rotating machinery. It has higher reliability compared with existing methods.