FRFT循频滤波及齿轮微弱故障特征提取

FRFT Filtering based on frequency tracking and extraction of gear's weak fault features

提出了一种分数阶傅里叶变换(Fractional Fourier Transform,FRFT)循频滤波方法,贴近瞬变工况下信号频率曲线变化特征,循迹剥离包含故障信息的特征分量,提取齿轮早期故障微弱特征。首先,研究了线性多尺度分段方法,将频率呈曲线任意变化信号自适应分成若干个频率近线性变化的信号段;然后研究了频率拟合确定FRFT滤波参数的方法,计算各段信号的FRFT滤波参数并逐段进行FRFT滤波,实现FRFT循频滤波。采用该方法对变速器加减速过程振动信号进行滤波解调分析,试验结果表明:线性多尺度分段方法,能自适应地将任一频率呈曲线任意变化信号分段成若干个频率近线性变化的信号段,且分段数较少;频率拟合确定FRFT滤波参数方法,不受振源和多分量数量影响,能准确确定各分段信号的FRFT滤波参数;该滤波方法能从变速器瞬变工况振动信号中循频提取出包含故障信息的特征分量,有效剥离其他分量和噪声干扰,对提取后的特征分量进行解调分析,能准确提取出传统方法难以识别的齿轮早期故障微弱特征。

To extract weak fault features of gears,a fractional fourier transformation （FRFT）filtering method based on frequency tracking （FT） was proposed to peel the feature components containing fault information by tracking frequency and to approach varying features of a signal＇s frequency curve under the transient condition.Firstly,the linear multi-scale segmentation method was studied to divide a signal frequency curve into some segments with frequency varying linearly.Then,the method to determine a FRFT filtering parameters with frequency fitting was studied to calculate FRFT filtering parameters of each signal segment,and each signal segment was filtered with FRFT filtering using these parameters.A gearbox＇s acceleration and deceleration vibration signals were filtered with FFFT and the filtered signals were demodulated.The results show that the linear multi-scale segmentation method is able to adoptively divide any signal frequency curve into several segments with frequency varying linearly;the FRFT filtering parameters determined with frequency fitting of each signal segment have no influences of vibration source and number of multi-component;the feature components containing fault information in gearbox＇s transient signals can be extracted with FRFT,the other components and noise are peeled at the same time;the weak features of the early fault of gear can be extracted through the demodulation analysis of the extracted feature components,they are difficult to be identified with the traditional method.