基于EMD-AR模型的齿轮箱故障诊断

Fault Diagnosis for Gearbox Based on EMD Method and AR Model

对基于EMD-AR模型的齿轮箱故障诊断进行了研究。对齿轮箱故障振动信号采用EMD方法进行分解,得到有限个平稳的IMF(本征模式函数),对其建立AR模型,将建立的每个AR模型残差的方差和自回归参数建立Mahalanobis距离判别函数,最后进行模式特征综合,确定齿轮箱的工作状态以及故障类型。研究表明,采用EMD-AR模型进行齿轮箱故障诊断是可行有效的,提高了齿轮箱故障检测的准确性。

A fault diagnosis for gearbox based on EMD method and AR model is studied. By using the EMD method fault signal of gearbox is decomposed to get a number of intrinsic mode function, then the AR model of each IMF is established. The auto-regressive parameters and variance of remnant of the AR model are used to obtain distance criterion function, and then integrate mode characteristics, by which to identify condition and fault pattern of gearbox. The experimental results show that the EMD method and AR model can effectively diagnose gearbox fault.