基于HHT和WNN的齿轮箱故障诊断

发布时间：2018-01-02 14:12

本文关键词：基于HHT和WNN的齿轮箱故障诊断　出处：《武汉工程大学》2011年硕士论文　论文类型：学位论文

【摘要】：近几十年来,由于齿轮箱故障诊断的至关重要而被广泛研究。本文是基于希尔伯特-黄变换(Hilbert-Huang Transform,简称HHT)和小波神经网络(Wavelet Neural Network,简称WNN)的齿轮箱故障诊断,Hilbert-Huang变换是一种新的时频分析方法,同时也是一种自适应信号处理方法。它包括经验模态分解(Empirical Mode Decomposition,简称EMD)方法和Hilbert谱分析两个过程。EMD方法是基于信号的局部特征时间尺度,将复杂的信号分解为有限的本征模态函数(Intrinsic Mode Function,简称IMF)之和;对本征模态函数应用Hilbert变换可以得到故障信号的Hilbert谱和Hilbert边际谱,从而能有效的提取故障特征,识别故障模式,进行故障诊断,这种自适应的分解方法非常适合于非线性和非平稳过程的分析。基于齿轮箱故障振动信号所表现的非线性非平稳特征,为了提取齿轮箱中的故障特征信息,本文首先对齿轮箱中采集的振动信号作小波包分解,对信号作降噪处理,同时选取特定频带的小波重构信号应用Hilbert-Huang变换进行了分析,得到经验模态分解(EMD)过程和一系列本征模态函数(IMF),选择特定的本征模态函数作Hilbert变换,获取振动信号的Hilbert谱和Hilbert边际谱,提取故障特征频率,有效的识别了齿轮箱中齿轮裂纹的不同故障模式。本文还提出了一种基于混合特征提取和WNN的齿轮箱故障诊断方法,采用时域分析法、小波分解和小波包分解相结合的方法对齿轮箱振动信号进行故障特征提取,将所提取的特征值作为WNN分类器的特征输入参数,采用反向传播(BP)算法对WNN结构中的平移参数、尺度参数、连接权值和阈值进行调整和优化。通过对三种具有不同裂纹尺寸的故障齿轮进行识别和分类,表明WNN有很好的模式识别和分类能力,能很好地应用于旋转机械的故障诊断。
[Abstract]:In the last several ten years, the gearbox fault diagnosis has been widely studied. This paper is based on Hilbert-Huang Transform, which is based on Hilbert-Huang transform. The gearbox fault diagnosis based on HHT and wavelet Neural network. Hilbert-Huang transform is a new time-frequency analysis method. It is also an adaptive signal processing method, which includes empirical Mode Decomposition. EMD method and Hilbert spectrum analysis are two processes. EMD method is based on the local feature of the signal time scale. The complex signal is decomposed into the sum of Intrinsic Mode function (IMF). The Hilbert spectrum and the Hilbert marginal spectrum of the fault signal can be obtained by applying the Hilbert transform to the intrinsic mode function, which can effectively extract the fault features and identify the fault mode. For fault diagnosis, this adaptive decomposition method is very suitable for nonlinear and nonstationary process analysis. Based on the nonlinear non-stationary characteristic of the gearbox fault vibration signal, in order to extract the fault characteristic information from the gearbox, the vibration signal collected in the gearbox is decomposed by wavelet packet. The signal is de-noised, and the wavelet reconstruction signal with special frequency band is analyzed by Hilbert-Huang transform. The empirical mode decomposition (EMD) process and a series of intrinsic mode functions are obtained, and the specific eigenmode functions are selected for Hilbert transformation. The Hilbert spectrum and Hilbert marginal spectrum of vibration signal are obtained, the fault characteristic frequency is extracted, and the different fault modes of gear crack in gear box are effectively identified. A method of gearbox fault diagnosis based on hybrid feature extraction and WNN is presented in this paper. Wavelet decomposition and wavelet packet decomposition are combined to extract the fault features of the gearbox vibration signal. The extracted eigenvalues are used as the input parameters of the WNN classifier. The translation parameter and scale parameter in WNN structure are analyzed by back-propagation algorithm. The connection weights and thresholds are adjusted and optimized. Through the identification and classification of three kinds of fault gears with different crack sizes, it is shown that WNN has a good capability of pattern recognition and classification. It can be used in fault diagnosis of rotating machinery.
【学位授予单位】：武汉工程大学
【学位级别】：硕士
【学位授予年份】：2011
【分类号】：TH165.3

【参考文献】