HHT与Elman神经网络在离心泵故障振动信号处理中的应用

Application of Hilbert-huang Transform and Elman Neural Network in Vibration Signals Processing for Centrifugal Pump Failure

提出了一种Hilbert-Huang变换(HHT)和Elman神经网络相结合的离心泵振动信号故障诊断新方法。首先,将离心泵振动信号时间序列数据经验模态分解(Empirical Mode Decomposition,简称EMD),然后经过Hilbert-Huang变换获得各模态(Intrinsic Mode Functions,简称IMF)的能量,并以“能量比”为元素,构造离心泵振动信号的特征向量,根据Elman神经网络模型能够逼近任意非线性函数的特点和具有反映系统动态特性的能力,利用Elman神经网络模型实现离心泵故障的诊断。实验研究结果表明该方法对离心泵振动信号故障具有很高的诊断率。

A new method of vibration signals fault diagnosis for Centrifugal Pump using the Hilbert-Huang Transform （HHT） combined with Elman neural networks was put forward. First, the series dat＇as of vibration signals for Centrifugal Pump were separated to components with different intrinsic mode function （IMF） by using empirical mode decomposition （ EMD）, then the Hilbert transformation was applied to every IMFs. The result of＇the method is the energy of every IMFs. The conception of ＂energy ratio＂ is proposed, based on the theory that signals energy in all IMFs can be affected by faults deeply, to construct feature vectors of Centrifugal Pump vibration signals. Based on the fact that the Elman model of neural network can well approach any nonlinear continuous function and has ability to reflect dynamic features of the systems. A Elman model is used to realize fault diagnosis for Centrifugal Pump. The experimental result indicates that this method, can show high diagnosis precision for the fault diagnosis of the Centrifugal Pump.