MOMED和双谱熵在液压泵退化特征提取中的应用

Application of MOMED and bispectral analysis to degradation feature extraction for hydraulic pumps

针对液压泵振动信号通常具有非线性强和信噪比低的特点,提出了一种基于多点最优最小熵解卷积(Multi point Optimal Minimum Entropy Deconvolution, MOMED)和双谱熵(Bispectral Entropy)的液压泵退化特征提取方法.首先针对最小熵解卷积(Minimum Entropy Deconvolution, MED)降噪效果受滤波器长度和迭代次数影响的问题,提出了一种多点最优最小熵解卷积(MOMED)降噪方法,并利用MOMED对液压泵原始振动信号进行处理,以降低原始信号中干扰成分的影响;然后采用双谱分析提取双谱熵作为退化特征,以提高对液压泵退化状态的反映能力;最后,通过对液压泵性能退化试验实测振动信号的应用分析,验证了该方法的有效性.

To solve the problem that the vibration signals of hydraulic pump usually appear with nonlinear and low signal to noise ratio, this paper presents a degradation state identification method based on Multipoint Optimal Minimum Entropy Deconvolution (MOMED) and Bispectral Entropy, First of all, an improved MOMED based on the slip window technology was proposed to overcome the problem of data partitioning. And then the improved MOMED was used to calculate four kinds of multi-fractal parameters of hydraulic pump. Five parameters are analyzed and as a result the singular index and the mean-rever-ting of generalized Hurst index are finally selected as the degradation feature.Finally,by analyzing actual data, results show that the proposed method can recognize the degradation status of hydraulic pump effectively.