基于Volterra级数和冗余提升小波包的机械隐含故障特征提取

Feature Extraction of Incipient Fault Based on Volterra and ULSP

针对提取机械原始振动信号中的隐含故障特征,提出了一种结合Volterra级数及冗余提升小波包(ULSP)的信号处理方法。先用二阶Volterra模型对信号进行延拓、预测,然后用冗余提升小波包对信号进行分解。对仿真信号的处理结果表明:分解得到的信号在边界没有振荡,有利于微弱特征的提取。工程应用中,完整地提取出了往复注水泵活塞与液缸密封碰磨产生的微弱故障特征信号。

In order to extract the incipient feature of vibration signal of machinery, a signal processing method combining with Voherra and undecimated lifting scheme packet （ULSP） is put forward. The two ends of original signal are extended and predicted by using the second-order Voherra series model. Then the extended signal is de-composed with the method of ULSP. The effectiveness of the proposed method is validated with simulated data be-cause of the thoroughly elimination of end distortion. Furthermore, in the processing of vibration signal of pump, the weak feature caused by the friction between plunger and cylinder is extracted and the incipient fault is detected.