基于自适应小波神经网络的往复泵故障诊断  

Fault Diagnosis of Reciprocating Pumps Based on Adaptive Wavelet Neural Network

作　　者：赵志华[1] 吴力[2] 刘继承 路敬祎[1] 

机构地区：[1]东北石油大学电气信息工程学院,黑龙江大庆163318 [2]大庆油田天然气分公司培训中心,黑龙江大庆163412

出　　处：《化工自动化及仪表》2013年第4期457-459,463,共4页Control and Instruments in Chemical Industry

摘　　要：为了对往复泵的故障进行正确诊断,提出了基于自适应小波神经网络的往复泵故障诊断方法。以往复泵单个泵缸内的压力信号作为系统特征信号采用小波变换来提取故障特征向量,同时将此特征向量作为小波神经网络的输入,利用小波神经网络对故障做进一步的精确实时诊断。通过对往复泵液力端多故障诊断实例的检验表明:该系统故障诊断正确率达到了93%以上。Considering accurately identifying the reciprocating pump＇ s fault types,a method based on adaptive wavelet neural network was proposed,which takes the pressure signal of reciprocating pump＇ s cylinder as the signal of system characteristics and extracts the fault feature vector through the wavelet transform,and meanwhile,it takes this feature vector as the wavelet neural network＇ s input so as to distinguish fault types at real time.Inspecting diagnosis instances of multiple faults of the reciprocating pump＇ s fluid end shows that the diagnostic accuracy rate can exceed 93%.

关 键 词：小波神经网络 往复泵 故障诊断 

分 类 号：TH165.3[机械工程—机械制造及自动化]