Fault Diagnosis Technology of Reciprocating Pumps based on Inlay Model Wavelet Neural Network

Abstract

2. Training Center, Natural gas branch of Daqing Oilfield, Daqing 163412 ChinaAbstract: A method is proposed based on inlay model Wavelet Neural Network in order to determine the reciprocating pump fault type accurately. This paper used the reciprocating pump single cylinder pressure signals as the characteristics of the system signal by wavelet packet decomposition to extract fault features vector, which is the input of wavelet neural network, and at the same time, used wavelet neural network to determine the type of the fault. Diagnosis of faults of fluid end on a reciprocating pump proves the system fault diagnosis accuracy to 94%.