抽油泵系统故障分类模型及其可视化终端的搭建  

The Construction of Fault Classification Model and Visual Terminal of Pump System

作　　者：邢鑫 吴伟[1] 魏航信[1] 李博[1] Xing Xin;Wu Wei;Wei Hangxin;Li Bo(School of Mechanical Engineering,Xi'an Shiyou University,Xi'an 710065,China)

机构地区：[1]西安石油大学机械工程学院,西安710065

出　　处：《机电工程技术》2023年第2期29-33,共5页Mechanical & Electrical Engineering Technology

基　　金：国家自然基金青年科学基金项目(编号:51405385);陕西省科技厅科技攻关项目(编号:2014K07-20)。

摘　　要：在石油开采行业中,及时掌握抽油泵系统的运行状态可以有效提高系统的运行效率和设备安全性。但是由于油井区域分布广泛且道路不便,所以油井的管理、维护成本高企。示功图作为抽油泵系统运行状态的直观表达,利用示功图图像作为样本构建的抽油泵系统故障分类模型是当前的主要研究方向。利用Pytorch搭建卷积神经网络模型实现示功图图像的分类,最终模型对抽油泵系统主要故障的分类准确率达97.92%。此外,处于不同区域的油井的当前运行状态通过4G模块将信息传输到由第三方开源库Kivy搭建的可视化终端进行显示。同时将该数据保存到MySQL数据库中,为后期抽油泵系统的管理和维修提供数据支撑。提出的抽油泵系统故障分类模型和可视化终端的设计方案可以有效地实现油井的实时监控、远程管理和高效维护,对后续构建智慧化油田有一定的借鉴和指导意义。In the oil extraction industry,keeping track of the operating status of the pumping system can effectively improve the operating efficiency and equipment safety of the system.However,due to the extensive distribution of oil wells and incommodious roads,the management and maintenance costs of oil wells are high.As a visual representation of the operating status of the pumping system,a fault classification model of the pumping system using the indicator diagrams as the sample is the main research direction at present.The Pytorch was usd to build a convolutional neural network model for classifying indicator diagrams,finally,the model had a classification accuracy of 97.92%for the main faults of the pump system.In addition,the current operating status of oil wells in various regions was transmitted to a visualization terminal built by a third-party open-source library Kivy through a 4G module for revealing the information.The data was also saved to a MySQL database to provide data support for later management and maintenance of the pumping system.The fault classification model of the pumping system and the design of the visualization terminal proposed could effectively realize real-time monitoring,remote management and efficient maintenance of oil wells,which could be a reference and guidance for the subsequent construction of intelligent oil fields.

关 键 词：抽油泵系统 示功图 卷积神经网络 4G Kivy 

分 类 号：TE933[石油与天然气工程—石油机械设备]