基于小波包-BP神经网络的输油泵机组转子故障诊断

Fault Diagnosis of Rotor Oil Transfer Pump Unit Based on Wavelet Packet-BP Neural Network

机械设备一旦发生故障,其输出信号的各频带成分就会改变,具体表现为对某些频带信号进行抑制,对另一些频带信号进行增强。不同的故障即使在相同的频带内,其信号所含的能量差别也比较大。为了利用这一现象作故障诊断,把频带能量的大小作为故障的特征参数,建立能量到故障的映射关系,进行故障的诊断。通过小波包分析提取振动信号各频带特征能量值,作为BP神经网络的输入,对泵机组转子系统故障进行诊断,结果显示,能取得较好的诊断结果。

In the event of failure of mechanical equipment,each band component of the output signal will change,and the specific performance is the certain band signal suppression and band signal enhancement for others. Different fault even within the same frequency band has the difference between the energy con-tained in the signal which is relatively large. Therefore,you can take advantage of this phenomenon,and take the size of the band energy as the characteristic parameter of failure,and a failure to establish the mapping relationship between energy was built to diagnose faults. This paper analyzed the vibration signal extracted by the wavelet packet characteristic energy value of each band,and as the input of BP neural network to have fault diagnosis of the pump unit rotor system. The results show that it can achieve better diagnosis results.