Research On Fault Diagnosis Method Of Pump Stuck In Mechanical Production Well Based On Indicator Diagram

With the development of oil field exploitation,the exploitation environment becomes more complex.The pump works hundreds of meters underground,which easily leads to pump stuck.At present,the fault diagnosis of the stuck pump mainly depends on the analysis of the indicator diagram by the monitoring personnel using the existing experience knowledge,which leads to the low efficiency of the fault diagnosis of the stuck pump and the heavy task of the staff.The traditional fault diagnosis method of stuck pump can not meet the development of oilfield intelligence,so it is very important to realize the intelligent diagnosis of stuck pump fault.In recent years,the development of computer communication and artificial intelligence technology has provided a new solution for the fault diagnosis of stuck pump in mechanized production well.In this paper,based on the research background of intelligent fault diagnosis method of stuck pump in mechanized production well,a fault diagnosis method of stuck pump in mechanized production well is proposed based on the analysis of fault diagnosis technology and indicator diagram at home and abroad,which provides effective service for the fault diagnosis and analysis of stuck pump.In order to apply the artificial intelligence method to the fault diagnosis of stuck pump,the main research contents of this paper are as follows:1.Research on the theory of indicator diagram and intelligent method of fault diagnosis: focus on the principle of indicator diagram diagnosis,the workflow of diagnosis and related intelligent diagnosis algorithm,analyze the feasibility and advantages of intelligent technology in the application of stuck pump fault diagnosis,design the fault diagnosis model of stuck pump based on indicator diagram,and provide a new method for fault diagnosis of stuck pump.2.According to the characteristics of the indicator data,a more comprehensive feature extraction method is adopted,which combines the moment feature extracted by Hu moment and the compression feature extracted by vector data compression algorithm to extract the load change information,so as to better reflect the difference of different card pump faults in classification diagnosis.3.On the classifier,select the support vector machine(SVM)which has better support for small samples,and use the fruit fly optimization algorithm(FOA)to optimize the kernel function parameters and penalty coefficients of SVM,and use the self-adaptive step length to improve the optimization algorithm of Drosophila,so as to make the algorithm more efficient.Finally,Ada Boost algorithm is used to strengthen the foa-svm classifier.Through the continuous iteration of Ada Boost algorithm,the weight value of all kinds of card pump fault types in the foa-svm classifier is balanced,avoiding the problem of high misdiagnosis rate for a certain fault.4.Based on the daily management of the card pump fault diagnosis business and the card pump fault diagnosis model,the card pump fault diagnosis system is designed and developed for the purpose of improving the intelligent level of the card pump fault diagnosis.The system's overall design,module function,key technology and other aspects are described in detail,and through the test and analysis of real environment data,the system performs well in the card pump fault diagnosis.The results show that the fault diagnosis method based on indicator diagram can make up for the problem of low intelligent level of pump stuck fault,which has important theoretical significance and practical value for the research of pump stuck fault diagnosis method.