基于延时自相关与重排Gabor变换的特征提取技术

Feature extraction based on delayed autocorrelation function and reassignment Gabor transform

根据自相关函数自身特性及时频分析重排算法,提出了基于延时自相关与重排Gabor变换的故障特征提取方法。本文首先利用时延自相关函数对信号进行降噪预处理,再对剔除干扰部分后的自相关函数进行重排Gabor变换,从而提取出故障特征频率,判定其故障类别。数值模拟仿真与齿轮箱故障试验结果表明:该方法更能有效地抑制噪声,凸显故障特征信息。因此,在旋转机械故障诊断领域,该方法具有广泛的应用前景。

According to delayed autocorrelation function and reassignment Gabor transform,a new method of feature extraction is proposed.In this paper,the noise signal is pretreated by delayed autocorrelation function,and then the autocorrelation function after eliminating the interference part is rearranged by reassignment Gabor transform,and the fault feature frequency is extracted and the fault category is determined.The simulation demonstrate that fault diagnosis based on delayed autocorrelation function and reassignment Gabor transform is effective in signal extracting and denoising,and could provides good prospects of application in failure analysis of rotating machinery.