液压泵信息融合故障诊断

Study on Fault Diagnosis of Data Fussion in Hydraulic Pump

针对液压泵故障特征的分散性和模糊性,提出基于振动和压力传感器的信息融合故障诊断方法。在充分分析液压泵球头松动故障机理的基础上,对振动信号和压力信号进行小波消噪处理,有效提取球头松动的故障特征。将不同类型特征参数进行特征层融合,利用主成分分析和改进算法的BP神经网络实现液压泵球头松动故障诊断。试验表明,基于不同类型传感器信息融合故障诊断方法可以有效地实现液压泵微弱故障的诊断。

Directing to the dispersiveness and faintness of failure characteristics of hydraulic pump, this paper presented the fault diagnosis method based on data fussions through measuring the vibration signals and pressure signals. On the basis of failure mechanism analysis of slipper loosing, this paper executed the wavelet analysis to cancel the interference existed in the measured signals. In order to improve the efficiency of fault diagnosis, the paper utilized Principal Component Analysis (PCA) and detective verification to decouple the failure eigenvectors. Inducting the reset eigenvectors into BP neural network, this work realized the multiple failure diagnosis with improved algorithm. The experimental results indicate that the method is content.