基于振幅熵和功率谱重心的转子振动故障诊断

摘要

对信号进行特征提取是故障诊断的关键,为了提高转子振动故障诊断的准确性,根据转子振动的特点提出了基于振幅熵H(A)与功率谱重心C的转子振动故障诊断方法.通过计算功率谱的重心得到表征功率谱变化的功率谱重心特征,计算振幅的熵值得到反映幅值分布特征与振动集中程度的振幅熵特征,组成二维特征量(H(A),C).然后通过转子故障模拟实验采集数据,对其进行DBSCAN聚类、K均值聚类、层次聚类、网格聚类4种聚类分析.结果表明,基于振幅熵H(A)与功率谱重心C的二维特征量(H(A),C)能够作为评价转子振动状态的综合特征指标.通过对传统的二维特征量(偏度、均方根值)、(裕度、标准差)运用网格聚类法进行转子振动故障诊断识别,结果表明,(H(A),C)的选取较于传统特征量的选取能更好地对转子运行中出现的常见故障进行区分.%Withdrawing the signal characteristic is a key to fault diagnosis.In order to improve the accuracy of the rotor vibration fault diagnosis,the fault diagnosis of rotor vibration based on amplitude entropy H(A)and power spectral centroid C was put forward by the feature of rotor vibration.The power spectral centroid C which expresses the changes of the power spectrum was obtained by calculating the center of gravity of the power spectrum,and the amplitude entropy which reflect amplitude distribution and vibration concentration was got by calculating the amplitude of entropy,then a two-dimensional characteristic (H (A),C)was presented.The clustering of DBSCAN,K-Means,hierarchical,grid were done with the data got by the rotor fault simulating experiment,and the results show that the two-dimensional characteristic (H(A),C)based on amplitude entropy H(A)and power spectral centroid C can be regard as comprehensive characteristic index to evaluate the vibration state of the rotor.And the fault diagnosis and identification of the rotor vibration was performed by doing the grid clustering analysis with the traditional two-dimensional characteristic (skewness,root mean square value),(margin,standard deviation),the results show that compared with the traditional characteristic,several common faults of the rotating machinery rotor can be better distinguished by choosing the two-dimensional characteristic(H(A),C).