摘要
齿轮故障振动信号具有多分量和调幅-调频等特点,导致振动信号耦合程度高、数据特征提取和识别难度大。提出了一种基于迭代经验小波变换(EWT)和稀疏滤波(SF)的振动信号故障特征提取和诊断方法。首先,利用尺度空间表示将齿轮振动信号的Fourier频谱自适应的划分为若干频带,并利用EWT将输入信号分解为若干本征模态函数(IMF);其次,利用互信息能量熵方法迭代去除振动信号中的噪声干扰成分,并重构振动信号;再次,建立基于稀疏滤波的无监督神经网络模型,将重构的振动信号作为神经网络模型的输入并学习故障特征,利用softmax辨识故障信息;最后,利用建立的故障诊断模型辨识齿轮故障测试数据并验证本文方法的有效性。结果表明,所提方法能够有效辨识故障特征。
Gear fault vibration signal has the characteristics of multi-component and amplitude-frequency modulation, which leads to high coupling degree of vibration signal and difficulty in extracting and identifying the data feature. A fault feature extraction and diagnosis method for vibration signal based on iterative empirical wavelet transform(EWT) and sparse filter(SF) is proposed in this paper. First, the Fourier spectrum of the gear vibration signal is adaptively divided into several frequency bands by using the scale space representation. The input signal is decomposed into several intrinsic mode functions(IMF) by the EWT method. Secondly, the mutual information energy entropy method is used to iteratively remove the noise interference components from the vibration signal and reconstruct the vibration signal. Thirdly, an unsupervised neural network model based on sparse filtering is established. The reconstructed vibration signal is used as the input of the neural network model and learned the fault features. The fault information is identified by the softmax. Finally, the fault diagnosis model is used to identify the test data of the gear fault and verify the effectiveness of this proposed method. The results show that the proposed method can effectively identify the different fault characteristics.
作者
辛玉
李舜酩
王金瑞
易朋兴
刘颉
Xin Yu;Li Shunming;Wang Jinrui;Yi Pengxing;Liu Jie(Nanjing University of Aeronautics and Astronautics,Nanjing 210016,China;Huazhong University of Science and Technology,Wuhan 430074,China)
出处
《仪器仪表学报》
EI
CAS
CSCD
北大核心
2018年第11期79-86,共8页
Chinese Journal of Scientific Instrument
基金
国家自然科学基金(51675262)
国家重点研究和发展项目(2016YDF0700800)
预研领域基金课题(6140210020102)项目资助
关键词
齿轮故障
迭代经验小波变换
频谱边界
稀疏滤波
无监督特征提取
gear fault
iterative empirical wavelet transform(EWT)
spectral boundary
sparse filtering
unsupervised feature extraction
