EMD与ICA相结合的复杂转子系统早期故障诊断

为了提取复杂转子系统微弱故障信息,对其早期故障进行预知诊断,针对某型涡桨发动机的减速器传动机构接连发生的齿轮毂裂纹故障问题,通过布置多组加速度传感器对多组正常齿轮毂和预制早期裂纹故障的齿轮毂进行正常试车下的振动信号采集。采用EMD(empirical mode decomposition)方法把测试信号分解成多个IMF分量,选取合适IMF分量利用基于非高斯性极大的ICA(independent component analysis)固定点算法进行混合再分离,得到了信息较为独立的特征分量。通过对特征分量进行解调分析得到能清晰反应故障状态的调制信号信息。结果表明基于EMD与ICA相结合的特征信号分离提取技术加包络解调法能有效地识别复杂转子系统早期故障信息。

Early Fault Diagnosis of Complex Rotor Systems by EMD-ICA

In order to extract the weak fault information of complex rotor system, and then make predict and diagnosis for early failure, aiming at the crack fault of gear hub which occurred repeatedly in reducers of one type of turboprop engines, arranging several acceleration sensors on the engine for collecting the vibration signal in the state of both normal and prefabricated early fault gear hubs under engine test. First decompose the test signal into several imf components using EMD method, then select appropriate imf components and mixed-separate them using ICA method which based on maximum non-Gaussian and Fixed-point, doing this we get the independent feature components. We could get the modulated signal information which can clear indicate the fault by doing demodulation of the independent feature components. The results shows that the characteristic signal separation and extraction technologies which based on combined EMD-ICA and envelope demodulation additional can effectively identify early fault information of complex rotor system .