摘要
在引入双谱、双相干谱和人工神经网络的基本理论之后 ,深入研究了汽车变速箱齿轮振动信号的双谱和双相干谱特征以及频率间的耦合情况 ,形成故障特征向量 ,并利用 BP人工神经网络 ,成功地将齿轮正常信号、磨损信号和断齿信号进行了分类。研究表明 ,对于齿轮振动信号 ,双谱比双相干谱更利于故障特征提取 ,不同状态下的齿轮振动信号中均存在二次相位耦合现象 ,但是耦合的形式不同 ,同时也表明 。
Bispectral analysis is emerging as a new powerful technique in signal processing, offering insight into non linear coupling between frequencies and having potential applications in many areas where traditional power spectral analysis provides insufficient information. In this paper, a new method based on bispectrum and artificial neural network for feature extraction and fault diagnosis of gearbox vibration signal is developed. Different gear faults can be identified successfully using this approach. Experimental results show that: (1)the bispectrum has an advantage over the bicoherence for feature extraction, (2)the quadratic phase coupling exists in all of the gear fault conditions, but distinct in kind, (3)the fault diagnostic methodology based on bispectrum and artificial neural network is effective.
出处
《振动工程学报》
EI
CSCD
北大核心
2002年第3期354-358,共5页
Journal of Vibration Engineering
关键词
故障诊断
双谱
特征提取
人工神经网络
齿轮故障
fault diagnosis
bispectrum
feature extraction
artificial neural network
