基于Hilbert-Huang变换的离心泵故障诊断

Centrifugal Pump Fault Diagnosis Based on Hilbert-Huang Transform

结合Hilbert-Huang变换(HHT)和径向基(RBF)神经网络的优点,提出将二者运用于离心泵故障诊断的新方法。利用HHT构造出代表离心泵振动信号的能量和频率分布的特征向量;根据RBF神经网络建立了从能量和频率分布的特征向量到故障模式的映射来实现离心泵故障诊断,对于离心泵的正常状态、转子不平衡、转子不对中、基础松动和油膜涡动及振荡故障具有较高诊断率。研究结果表明,该方法可有效对离心泵振动信号进行诊断。

Combining the advantages of Hilbert-Huang Transform(HHT) and Radial Basis Function(RBF) neural networks, a new method for fault diagnosis of centrifugal pumps is proposed. The HHT is used to construct the eigenvectors of the energy-frequency distribution representing the vibration signal of the centrifugal pump. According to the RBF neural network, the mapping from the eigenvectors of the energy-frequency distribution to the fault mode is established to realize the centrifugal pump fault diagnosis, which has a high diagnostic rate in the pump’s normal state, rotor imbalance, rotor misalignment, foundation looseness and oil film whirl and oscillation faults. The research results show that this method can effectively diagnose the vibration signal of centrifugal pump.