Fault Diagnosis Of Fracturing Pump Based On Synchro-squeezed S Transform And Deep Residual Network

Petroleum has become an indispensable and important resource in the world.With the gradual increase in exploitation,the output of oil wells is also decreasing.Fracturing mining is one of the most effective methods to improve oil recovery,fracturing truck is the main equipment for fracturing construction,which plays an important role in reducing production cost and improving oil and gas recovery.The fracturing pump is the core part of the fracturing truck.The working state of the fracturing pump will have a direct impact on the working performance of the fracturing truck.Based on the changeable working conditions and poor working environment of the fracturing truck,the fracturing pump is prone to failure after a long time of operation.Once an accident occurs,it will cause serious losses.Therefore,it is of great significance to diagnose the fracturing pump of the fracturing truck in time.There are many excitation sources of fracturing pump,and the complexity of vibration signal makes the fault diagnosis more complex.In view of the complex vibration signal of fracturing pump and the difficulty of accurate extraction and identification of fault features,this paper studies the intelligent diagnosis method of fracturing pump of fracturing truck.The main contents are as follows:(1)The research status of fracturing truck fault diagnosis methods and future development trends are discussed,and the existing fault diagnosis methods of mechanical equipment are analyzed,based on this,the specific structure of the paper is given.(2)The working principle of the fracturing pump is analyzed,and the typical failure types of the fracturing pump are studied in detail.At the same time,the causes of wear between the key components of the fracturing pump are analyzed.(3)In view of the complex structure of the fracturing truck,the harsh working environment,and the vibration signal containing more noise,it is difficult to rely on the time domain waveform or frequency domain waveform of the signal to identify the state.Noise method.The research results show that this method can effectively remove the noise component,and the time-frequency resolution is higher than other methods.(4)Aiming at the problem that the accuracy of the Convolutional Neural Network tends to be saturated and it is impossible to obtain better training results by increasing the number of network layers,a classification and recognition model of the deep residual network DRN is constructed,combined with the superior time-frequency decomposition characteristics of Synchro-Squeezed S Transform.A fault diagnosis method for fracturing pumps of fracturing trucks based on Synchro-Squeezed S Transform SSST and Deep Residual Network DRN is proposed.The experimental results show that this method is free from the dependence on expert experience and can be effective in a strong noise environment Improve the accuracy of diagnosis of fracturing pumps.