基于最大相关峭度解卷积行星齿轮箱微弱故障诊断

Weak Fault Diagnosis of Planetary Gearboxes Based on Maximum Correlation Kurtosis Deconvolution

最小熵解卷积(MED)是一种常规的微弱故障特征提取方法,对局部故障脉冲有比较好的提取效果,但是对于含有周期性故障脉冲的振动信号,故障特征识别率比较低。微弱故障时候的行星齿轮箱产生的振动信号通常是周期性的,MED不能取得比较好的识别效果。针对行星齿轮微弱故障特征难以提取的问题,将最大相关峭度解卷积(MCKD)方法应用到行星齿轮箱微弱故障特征提取中。MCKD避免了最小熵解卷积对周期性冲击识别度低的缺点,同时可以有效抑制行星齿轮箱中谐波和噪声分量,准确地识别出行星齿轮箱所处状态。为了验证该方法在行星齿轮箱中的应用价值,将两种方法分别应用在传动系统综合诊断平台收集到的振动信号中,结果表明MCKD算法对于行星齿轮箱微弱故障识别有比较好的效果。

Minimum entropy deconvolution(MED) is a conventional weak fault feature extraction method, which has good extraction effect on local fault pulse. However, for the vibration signal containing periodic fault pulse, the fault feature recognition rate is relatively low. The vibration signal generated by the planetary gearbox with weak faults is usually periodic, and MED cannot achieve better recognition results. Aiming at this problem, this paper applies the maximum correlation kurtosis deconvolution(MCKD) method to the weak fault feature extraction of planetary gear boxes. MCKD can avoid the disadvantage of low recognition rate of periodic impact of minimum entropy deconvolution, and can effectively suppress harmonic wave and noise components in planetary gearboxes, and accurately identify the state of the planetary gearboxes. In order to verify the application value of this proposed method in planetary gearboxes, the two methods of MED and MCKD are applied to the analysis of the vibration signals collected by the comprehensive diagnosis platform of transmission systems.The results show that the MCKD algorithm has a better effect on the weak fault identification of planetary gear boxes than MED.