基于小波和灰色关联度的消除负频率干涉方法

对于机械振动与故障诊断领域中常见的低频滚振等信号,负频率干涉是影响其频谱分析校正精度的重要因素。为消除负频率干涉,提出了一种基于小波变换和灰色关联度的极低频信号频率分析校正新方法。首先将原信号构造为标准参考序列,然后对原信号进行一维小波变换,得到以变换尺度和采样点数为行数和列数的信号分量矩阵,计算变换后的信号分量矩阵中各行分量与标准参考序列之间的灰色关联度,得到灰色关联序列。由于负频率干涉分量的灰色关联度系数较小,可视为信号中的＂无用成分＂。以灰色关联度系数平均值作为阈值,将小于阈值的负频率干涉分量剔除,最后对剩余信号分量重构恢复出已消除负频率干涉后的原信号。仿真实验和工程实例验证均表明了所提方法的有效性。

New method to eliminate negative frequency interference based on wavelet transformation and grey correlation degree theory

Low frequency vibration signal is common in mechanical vibration and fault diagnosis domain.One of the most important factors that influence the accuracy of low frequency spectrum is the negative component intervenes.In order to eliminate such interference,a new method based on wavelet transformation and grey correlation degree were proposed.Firstly,the original signal was constructed as a standard reference arrays,then the wavelet transformation were carried out to gain subsection matrix relates to the scale of the wavelet function and the number of sample points.A grey correlation sequence is gained from the grey correlation degree coefficients that relates between each subsection in the row of the matrix and the original signal.The average of the degree coefficients was then calculated to judge the element acceptance or rejection of the subsection matrix.Negative component intervene subsection which can be regarded as the unnecessary composing were abandoned for its low grey correlation degree coefficients.The corresponding element of the subsection matrix was replaced by the value of zero.The remaining subsection was reconstructed to renew the original signal at last.The simulation experiments and the engineering test show the validity and efficiency of the proposed method.