| Reciprocating pump as the core equipment in the field of oil drilling, it has high pressure, large power and bad working environment characteristics. Especially when the pumping medium of reciprocating pump contains solid particles or other hone and corrosive impurity, the hydraulic side parts are easy to occur failure frequently, leading to the well homework is not completed smoothly, serious problems may even endanger the life property safety of attendants and cause great damage. At present, the hydraulic end performance monitoring and fault diagnosis of reciprocating pump has not a complete set test system for field, mostly relied on traditional diagnosis methods, such as experienced worker’s ear visual, to repair when the pump is shutdown.This method cannot judge the fault position and fault degree accurately, need to open it all and one by one for investigation. The workload is big and diagnosis efficiency is low. The current intelligent fault diagnosis methods are mainly composed of vibration signal test, but there are a lot of the fault vibration signal interference sources in the scene test and fault diagnosis method based on vibration test technology, the influence of the vibration source fault diagnosis model is complex, difficult to extract the fault feature information, it is difficult to efficiently and accurately to the realization of reciprocating pump hydraulic performance monitoring and fault diagnosis. In this paper, the fault diagnosis method based on the pump cylinder pressure test, not affected by the vibration source, can realise performance monitoring and fault diagnosis of reciprocating pump hydraulic end accurately and fastly.Firstly, based on the study of domestic and foreign reciprocating pump device performance monitoring and fault diagnosis method, this paper summaries the reciprocating pump common faults and their causes, and puts forward the method of fault diagnosis based on the pump cylinder pressure test. A hardware system for performance monitoring and fault diagnosis of the reciprocating pump is designed.Secondly, a set of fault parts of the hydraulic end of the reciprocating pump is processed, simulated suction valve spool, valve seat, spring, valve core, valve seat, spring and plunger 7 fault parts respectively, and 3 kinds of different fault degree, mild wear, moderate wear and severe wear. Through the 3DS-1/12.5 type high pressure plunger pump makes a large number of fault simulation test using the hardware test system, the normal state and the 19 fault states are simulated respectively, and analyzes the rule of fault indicator diagram.Then, select the 30 indicator diagram of various conditions, through the method of indicator diagram graphic feature extraction by moment feature and gray matrix of two feature extraction, got characteristic features of the 420 diagrams under 14 conditions. The training sample set of fault recognition is used as the first 20 groups of different working conditions, the last 10 groups of characteristics as a test sample. Support vector machine algorithm and biomimetic pattern recognition are used to identify two kinds of pattern recognition, to complete performance monitoring and fault diagnosis of the hydraulic end of reciprocating pump, the fault diagnostic efficiency as high as 92.3%.Finally, to facilitate the popularization and application of the diagnostic method, it uses MATLAB software program to complete the software system design of the performance monitoring and fault diagnosis system of the pump based on the pump cylinder pressure test. the software system has four functions:signal processing, performance monitoring, fault diagnosis and data storage. It can realize the automatic drawing of the pump cylinder indicator diagrams, extract indicator diagram graphic feature by moment feature, realize performance monitoring and fault diagnosis based on bionic pattern recognition method, and obtain the diagnosis results and the corresponding repair advice. |
PDF Full Text Request