基于自适应谐波小波和能量熵的转子系统故障诊断研究

Fault diagnosis of rotor system based on adaptive harmonic wavelet and energy entropy

针对转子系统非平稳振动时故障特征难以准确提取的问题,提出一种基于自适应谐波小波和能量熵的转子系统故障诊断方法。首先,采用连续谐波小波方法分解转子信号,克服"二进制"谐波小波包分解不能任意选取感兴趣频段的缺限,同时在分解过程中通过时间尺度变换的方式消除信号采集过程中不同转速及采样频率的影响;然后,通过设定合理的分解参数,提取出表征转子系统的故障特征信息并构建故障模式矩阵,得到转子系统早期局部碰摩、全周碰摩、油膜涡动和油膜振荡等4种工况下的能量熵值;最后,将特征向量输入支持向量机(support vector machine,SVM)判断出转子系统的故障类型。试验结果表明:该方法可以有效用于转子系统的故障诊断。

In order to solve the problem that fault feature of rotor system was difficult to accurately extract because of non-stationary vibration, a new method based on harmonic wavelet adaptive decomposition and energy entropy was presented in this paper. Firstly, continue harmonic wavelet was used to decompose the signal of rotor system, which broke the constraint that binary harmonic wavelet decomposition could not select any interested frequency band. Time scale transformation method was applied to the process of decomposition in order to eliminate the influence by different rotational speed and sampling frequency. Secondly, fault feature information of rotor system was extracted by setting reasonable parameters and fault pattern matrix was constructed, then energy entropies of rotor system under four working conditions were obtained. Finally, characteristic vectors were served as input vectors of support vector machine to identify fault patterns of rotor system. The result showed that the proposed method can diagnose fault of rotor system effectively.