一种基于冗余小波变换等效功率谱的故障特征提取方法

A fault feature extraction approach with equivalent power spectrum based on redundant wavelet transformation

为了提取信号中较弱的故障特征频率分量,以正交小波为基函数,在对振动信号实施冗余小波变换分解的基础上,提出了一种等效的功率谱分析方法。该方法利用冗余小波变换分解提供的多分辨信息,先剔除高频噪声分量,在获得其他尺度的功率谱后,对其进行归一化处理。由于每个分解频带的分析频率相同,因此叠加归一化的功率谱得到信号的等效功率谱,使被淹没信号中的微弱频率分量在功率谱上显现出来。采用这种方法有效地提取了烟汽轮机汽封与转子之间摩擦故障的特征信息。

In order to extract weak fault features from vibration signals efficiently,an equivalent power spectrum analysis approach was proposed based on the redundant wavelet transformaction(RWT)decomposition by using an orthogonal wavelet as a basis function.The new approach made full use of multi-resolution information obtained with the RWT decomposition.Firstly,noise component in high-frequency-band was omitted,and the other decomposition components were treated as useful components to calculate their power spectra,respectively.Then,a normalized operation was made for these power spectra.Because the analysis frequency band of the RWT decomposition components was the same,the equivalent power spectrum was gained by adding the normalized power spectra.As a result,some weak frequency components submerged in the signal before were revealed in the equivalent power spectrum.The proposed approach was used to diagnose friction fault between gas-seal and rotor of a gas turbine in an oil refinery and the fault features were detected effectively.