基于多尺度排列熵的液压泵故障识别

将排列熵引入液压泵的故障识别中,分析了排列熵作为液压泵故障特征指标的性能；采用互信息法和伪近邻法优选排列熵计算中的延迟时间和嵌入维数，基于优选参数得到了能够更好区分液压泵故障的排列熵。针对单尺度排列熵只能在单个尺度上衡量振动信号复杂度的不足,在对多尺度排列熵进行研究的基础上提出了一种综合多尺度排列熵熵值和排列熵变化趋势的指标--多尺度排列熵偏均值,对液压泵实测信号的分析结果验证了该指标作为液压泵故障特征的有效性和优越性。

Fault Identification of Hydraulic Pump Based on Multi-scale Permutation Entropy

A permutation entropy was introduced into the fault diagnosis procedure of hydraulic pump and its ability to distinguish different faults was analyzed herein.The mutual  information and false nearest neighbor algorithm were used to select the delay time and embedding dimension independently, the permutation entropy which was based on the selected parameters had more excellent performance to distinguish the faults of hydraulic pump. Aiming at the shortages of permutation entropy only measuring the complexity of the vibration signals in only one scale, the partial mean of multi-scale permutation entropy which composited the value of multi-scale permutation entropy and its variation tendency was proposed based on the study of multi-scale permutation entropy, and it was used as a parameter to distinguish the difference of various faults of pump. The results of actual testing vibration signals demonstrate that the proposed parameter is more efficient and more excellent than that of traditional PE to distinguish the faults of hydraulic pump.