Application of ontology to knowledge management of sucker rod pumping system fault diagnosis

The comprehensive diagnosis ontology database, including the knowledge domains of fault, collection and structure, is established. And the relationships among the ontology are described.
Nội dung trích xuất từ tài liệu:
Application of ontology to knowledge management of sucker rod pumping system fault diagnosisJournal of Automation and Control Engineering, Vol. 1, No. 2, June 2013Application of Ontology to KnowledgeManagement of Sucker-rod Pumping SystemFault DiagnosisPeng ChengTsinghua University, Beijing, PRCEmail: pengchengthu@foxmail.comYan JianXi’an Shiyou University, Shanxi Xi’an, PRCEmail: 704874024qq.comAbstract—Sucker-rod pumping system operating statusfault diagnosis technology is closely related to the energyconsumption, production per well and the cost of crude oilexploration. But the knowledge depiction of the expertdiagnostic system is mainly based on the production rules,neural networks or object-oriented. As ontology hasadvantages in hierarchy and semantic relationshipsexpressing, it is can be used in the knowledge managementin the expert diagnostic system of the fault diagnosistechnology. Then an ontology model for sucker-rodpumping system running state fault diagnosis system is alsoproposed in this paper. Basing on ontology, the knowledgemanagement system framework is built up, which includesdata layer, logic layer and application layer. Thecomprehensive diagnosis ontology database, including theknowledge domains of fault, collection and structure, isestablished. And the relationships among the ontology aredescribed. this technology appeared after the emerge of expertsystems [13], [14], [22], [23], neural networks [15] andpattern recognition technology [16]-[21].The expert diagnosis system developed by H. J. Derk[13] has three main auxiliary programs: API, DHOLEand VIEW. By aid of the auxiliary programs, some taskscan be done: carry out the designs for conventionalpumping device to get the rated operating parameterswhich are used as the reference data, calculate thehydrodynamic forces and inlet pressure of the pumpbasing on the transformation from the ground indicatordiagram into downhole one, and so on. It also can helpusers to find out one certain fault group similar to thedownhole indicator diagrams from all typical indicatordiagrams in various kinds of fault group to check allindicator diagrams belong to this fault group. Then userscan determine the fault category and get expert’s explainand corresponding countermeasure.The expert system developed by W. L. Foley et al. [14]is an auxiliary function module of the surface anddownhole analysis program (SADA) which is a computerdiagnostic technology. It includes a statistics-type modeldiagrams and a group of production rules featuring theform of If-then. They are built up according to theexpert experience and knowledge. The database has over100 types of model diagrams corresponding to differenttypical failures of the pumping system, as well as morethan 60 items of If-then rules of thumb, such as If thedownhole indicator diagram shows that the liquid levelimpact and pump inlet pressure is higher than the casingpressure, then it is indicated that the density of the liquidhas changed or pump diameter is inappropriate .The pumping system fault recognition softwaredeveloped by R. R. Dickinson et al. [16] established astandard indicator diagram library having about 30 kindsof corresponding typical pumping system faults. It isbased on the indicator diagrams features: boundary chaincode, boundary curve Fourier descriptors and boundarygray matrix. Incorporating pattern recognition techniquesand the criterion of minimum Euclidean distance, the userIndex Terms—Ontology, pumping system, fault, knowledgemanagementI.INTRODUCTIONSucker-rod pumping system (short of “pumpingsystem”) running state fault diagnosis technology hasbeen one of the important issues in oil productionengineering. This technology can affect the exploitationcost, energy consumption and the production per well.After 80 years of studies and practices, it has been greatlydeveloped, especially, with the emerge of computerdiagnostic technology, has entered into a new phase, i.e.it has developed from the qualitative analysis to thequantitative analysis [1], at the same time, it has steppedinto the intelligent diagnostic phase [2]. There are fourdevelop stages in pumping system fault diagnosistechnology: ground indicator diagram analysis [3]-[5],downhole indicator diagram diagnosis method [6],computer diagnostics [7-12] and artificial intelligencediagnostic method [13-23]. The practical applications ofManuscript received September 25, 2012; revised December 24,2012.©2013 Engineering and Technology Publishingdoi: 10.12720/joace.1.2.126-131126Journal of Automation and Control Engineering, Vol. 1, No. 2, June 2013can judge the operation status and diagnose, then give aproper countermeasure.Basing on the study of R. R. Dickinson et al., G. A. Yuet al. [17] established a standard indicator diagram librarywith 17 categories of typical pumping system faults. Theclassification and identification to the pumping systemoperation status by aid of the criterion of minimumEuclidean distance and absolute difference distance, thecorrect recognition rate is greatly improved. After furtherimprovements, through extraction of the statistical natureof the indicator diagram boundary grayscale matrix [18]and incorporating the sample average self-learningfunction and Fisher linear classifier [19], the recognitionspeed and accuracy was further improved. In present, thistechnology is widely used in oilfields of China.Above mentioned diagnosis technologies and softwareare efficacious for single pumping system fault and it hasverified an exciting economical benefit in increase crudeoil production, reduce energy consumption and saveworkover cost. However, as to these techniques includingexpert system, neural network or pattern recognition, it ishard to carry out the knowledge and informationacquisition, sensitivity and uncertainty reasoning, selflea ...