Research On Drilling Pump Fault Diagnosis System Based On Vibration Test

Drilling pump is an important equipment in oil drilling cycle, in the process of drilling due to bad conditions, often because of the fault led to the construction of stagnation, thus affecting the drilling process. Currently in petroleum assessment working condition of drilling pump depends mainly on the engineer experience, is generally the fault by ear, visual way to judge the drilling pump. This way can not be timely and accurately determine the pump failure, and the risk of job insecurity, often cause the pump efficiency is low, there is a serious fault, seriously affected the drilling process. Site need a test method to realize the field drilling pump test, in order to meet the needs of drilling pump working condition assessment. In this paper, the vibration test method of drilling pump collecting vibration signals, detection of drilling pump current running state, can carry out the maintenance and maintenance.Firstly, this thesis studies the domestic and foreign literatures and field drilling pump repair, analysis and summary of drilling pump failure mechanism and repair common failures, process. In order to distinguish and identify the severity of drilling pump the fault, the fault is divided into class I and class Ⅱ fault fault two kinds of drilling pump repair, whether immediate repair provides guidance and advice. In this paper, the vibration signal acquisition are filtering, time-domain analysis, spectrum analysis and wavelet packet decomposition technique.Secondly, this paper developed the hardware system of fault diagnosis of drilling pump. And according to the hardware system has been tested in the field. The experiment was divided into trouble free drilling pump optimal sensor location test, fault test and measured drilling pump test. The fault is the fault overhaul drilling pump parts are single installed in the fault free pump, collecting vibration signals corresponding, establishing fault database. The drilling pump test is to choose the same manufacturers similar fault pieces of the same type of drilling pump as the diagnosis of drilling pump test, verifying the accuracy of fault test signal.Then, this paper adopts two methods to identify the fault diagnosis based on the quantity of small Potter syndrome. A rough set BP neural network fault diagnosis based on the theory, using the theory of rough set reduction information system, and then use BP neural network fault diagnosis, and using field data and the actual fault is verified. The other is probabilistic neural network fault identification, introduces the source and the working principle of probabilistic neural network is verified by the data, the effect of recognition. By comparing the two kinds of fault identification method, compared the advantages and disadvantages of the two methods, and ultimately determine the quick, effective identification of drilling pump failure probability neural network.Finally, the fault diagnosis system of drilling pump software has been writed. The hybrid programming technology of MATLAB and VB, using the powerful operation function of MATLAB and VB friendly interface written, thereby reduced the workload of programming. It introduces the realization process of mixed programming, give full play to two software the advantages of MATLAB and VB. The module introduces the function and Realization of software design, the final completion of the main program drilling system fault diagnosis software of pump.