摘要
针对轴向柱塞式液压泵性能退化中振动信号非线性强、退化特征提取困难等问题,提出基于形态非抽样融合与DCT(Discrete Cosine Transform)高阶奇异熵的退化特征提取方法。在一般框架下提出形态非抽样小波融合方法,通过构建特征能量因子筛选各分解层近似信号,据融合规则实现双通道振动信号融合重构、改善重构信号的特征信息;并利用DCT高阶谱分析法对融合信号进一步处理,通过奇异值分解分别计算Shannon、Tsallis奇异熵作为液压泵性能退化特征向量;用仿真信号及液压泵实测振动信号验证该方法的有效性。
To solve the problem that vibration signals of a hydraulic pump as usual are strongly nonlinear and its degradation features are difficult to extract,a degradation feature extraction method based upon morphological undecimated wavelet decomposition fusion （MUWDF）and DCT high order singular entropy was proposed.The MUWDF algorithm was presented under the general framework of morphological undecimated decomposition. The approximate signals of all decomposition layers were selected by using the feature energy factor and dual-channel vibration signals were fused according to the presented fusion rules so as to increase the proportion of feature information.On this basis,a high order spectrum analysis algorithm modified by DCT was proposed for further dealing with the fused signal.Shannon and Tsallis singular entropies,which were considered as fault degradation features of hydraulic pump,were respectively achieved by singular value decomposition.Finally,the proposed method was verified by using simulation signals and real pump vibration signals in various working conditions.
出处
《振动与冲击》
EI
CSCD
北大核心
2015年第22期54-61,81,共9页
Journal of Vibration and Shock
基金
国家自然科学基金资助项目(51275524)
关键词
退化特征提取
形态非抽样小波融合
DCT
高阶奇异熵
degradation feature extraction
morphological undecimated wavelet decomposition fusion
DCT
highorder singular entropy
