Fault Diagnosis Of Electric Submersible Screw Pump System

Because of its simple wellhead device,wide range of transmission medium and high efficiency,the number of application of the submersible screw pump oil production system in major oil fields is increasing year by year.However,because the main parts of the submersible screw pump are mainly in the underground,it is difficult to directly determine the failure mode when the failure occurs,which seriously affects the life of the unit and the normal production of the oil well,and at the same time hinders the popularization and application of the submersible screw pump.In order to solve the problem that the failure form of the submersible screw pump is difficult to distinguish,this paper intends to carry out the fault diagnosis research of the submersible screw pump.Based on the analysis of the working characteristics of the submersible screw pump,this paper analyzes the influence of the changes of the working characteristics of the submersible screw pump on its working state,and puts forward two fault diagnosis methods applied to different situations of the submersible screw pump.The main contents are as follows:(1)Taking the production system of the submersible screw pump as the research object,this paper analyzes the common fault types,the movement characteristics,mechanical characteristics,hydraulic characteristics and efficiency characteristics of the screw pump,and finds out the relationship between the change of the working characteristics of the submersible screw pump and its working state.(2)A fault diagnosis method based on probabilistic neural network is proposed.The wavelet packet time-frequency analysis and probabilistic neural network are combined.The active power,output,oil pressure,casing pressure and liquid level depth are taken as the parameters to characterize the working state of the submersible screw pump,and the fault diagnosis model of the submersible screw pump is established.On the platform of Matlab GUI,the fault diagnosis software of submersible screw pump is developed.(3)A fault diagnosis method based on data mining for the production system of submersible screw pump is proposed.Using the statistical method and Hadoop big data processing platform,a random forest model is established.Nine parameters,including active power,submergence,daily liquid production,oil pressure,casing pressure,sand,wax,screw pump model and well number,are used as the working state characterization parameters to identify the working state of the screw pump.According to different data types,two fault diagnosis methods based on probabilistic neural network and data mining technology are formed in this paper,with the accuracy of 90.5% and 92.8% respectively.The two methods can effectively and accurately analyze and diagnose the fault of the submersible screw pump,so as to reduce the production loss of the oil well and promote the further application of the submersible screw pump in the oil field.