基于IWMMDF的液压泵振动信号特征提取方法研究

Research on Feature Extraction of Hydraulic Pump Vibration Signals Based on IWMMDF

根据液压泵振动信号的非线性及信噪比低等特点,针对现有加权方法存在的不足,提出了基于改进加权多尺度形态学差值滤波器(improved weighting multi—scale morphological d—value filter,IWMMDF)的液压泵振动信号特征提取方法;该方法首先利用不同尺度的结构元素对信号进行形态差值滤波,再根据本尺度特征能量比占所有尺度特征能量比总和的大小对各尺度的滤波结果进行加权求和;仿真实验和液压泵故障模拟试验结果表明,该方法不但避免单尺度形态分析的片面性,而且可获得不同层次上的信号特征信息,能够有效的提取液压泵振动信号的特征。

According to the nonlinear and signal--to--noise ratio lowly of hydraulic vibration signal, aiming at the deficiencies of the prior weighting methods, improved weighting multi--scale morphological D--value filter （IWMMDF） for feature extraction of hydraulic pump vibration signals is proved. Firstly, the method proved in this paper filtering signals with morphological D--Value by using the structural ele- ments. Second of all, weight all the scales results based on the ratio o~ this scale feature energy ratio （FER） in the sum of all the scales＇ FER. Through numerical simulations and hydraulic pump fault simulation tests, the method proposed avoiding slngle--scale morphological a nalysis of one--sidedness, and obtaining the information of signal characteristics in different levels at the same time, could be clearly extract the feature of hydraulic pump vibration signals.