Тип публикации: статья из журнала
Год издания: 2019
Идентификатор DOI: 10.5937/jaes17-22908
Ключевые слова: Classification, Electric centrifugal pump, Models, Oils, Recognition, Technical state, Technology
Аннотация: The article deals with the problem of constructing models for automation of the technological situations recognition procedure during operation of oil wells. An approach was suggested to recognize technological situations associated with the operation of electrical centrifugal pumping units in oil production. The paper describes the methods for constructing models intended to recognize technological situations characterizing different types of failures of such electric centrifugal pumping (ECP) units. The models based on artificial neural networks, classification trees and support vector machines were considered as separate methods for constructing models for recognizing the technical state of ECP units in oil production. The paper presents the results of studying such methods in the tasks of assessing the technical state of several types of oil and gas production equipment. It is proposed to use sets of models enabling to integrate solutions of individual recognizers to improve situation recognition reliability. In the course of the research, tests were carried out on real operational data of ECP units. The research results showed that the use of such complex models will ensure a sufficiently high accuracy of recognition of technological situations. The proposed complex models provide higher stability of the results, which is confirmed by the results of statistical analysis of solutions obtained in the course of numerical experiments. Thus, it is shown that the proposed complex models for recognition of technological situations are an effective option to be used in object control systems during operation of oil producing wells. © 2019 Institut za Istrazivanja. All rights reserved.
Журнал: Journal of Applied Engineering Science
Выпуск журнала: Vol. 17, Is. 4
Номера страниц: 541-549
ISSN журнала: 14514117
Издатель: Institut za Istrazivanja
Информация о публикациях загружается с сайта службы поддержки публикационной активности СФУ. Сообщите, если заметили неточности.