基于HHT变换和SVM的齿轮箱故障诊断

HHT Convert and SVM Based Gearbox Fault Dignosis

提出了基于HHT变换和SVM结合的齿轮箱故障诊断方法,介绍了固有模态函数、EMD分解和Hilbert谱以及支持向量机(SVM)理论。先对各种工况信号消噪,再利用EMD分解将信号分解为IMF分量,求出Hilbert谱和边际谱,再利用边际谱求出各信号的故障特征信息,最后利用支持向量机判别齿轮箱的故障类型。该实验证明了此方法诊断齿轮箱故障的有效性。

New gearbox faults diagnosis method is presented based on Hilbert-Huang convert and SVM. The theory of intrinsic mode function, EMD decomposition and Hilbert spectrum, as well as sup- port vector machine （SVM） are introduced. Firstly, a variety of conditons signals are de-nosied, second- ly, the signal is decomposed IMF components which make use of EMD decomposition,and Hilbert spectrum and marginal spectrum is obtained, thirdly, variety of signal fault information is obtained with marginal spectrum, finally, gearbox fault types are discriminated by support vector machine. The experiments proved the effectiveness of the method which diagnose gearbox faults.