查询结果:   段友祥,李钰,孙歧峰,徐冬胜.改进的Alexnet模型及在油井示功图分类中的应用[J].计算机应用与软件,2018,35(7):226 - 230,272.
中文标题
改进的Alexnet模型及在油井示功图分类中的应用
发表栏目
人工智能与识别
摘要点击数
976
英文标题
IMPROVED ALEXNET MODEL AND USING IN DYNAMOMETER CARD CLASSIFICATION
作 者
段友祥 李钰 孙歧峰 徐冬胜 Duan Youxiang Li Yu Sun Qifeng Xu Dongsheng
作者单位
中国石油大学(华东) 计算机与通信工程学院 山东 青岛 266580     
英文单位
College of Computer and Communication Engineering, China University of Petroleum (South East), Qingdao 266580, Shandong, China     
关键词
示功图 深度学习 卷积神经网络 Alexnet
Keywords
Dynamometer card Deep learning Convolutional neural networks Alexnet
基金项目
国家“十三五”重大专项资助项目(2017ZX05009-001)
作者资料
段友祥,教授,主研领域:人工智能,图形图像处理。李钰,硕士生。孙歧峰,讲师。徐冬胜,硕士生。 。
文章摘要
现在有杆抽油机采油设备仍在原油开采中占据主导地位,示功图采集及分析是检测、预防、解决采油生产过程中各种故障的有效措施和手段。借助人工智能方法进行油井抽油机示功图自动分类识别和故障判断一直是研究的重点。深度学习为示功图识别和解释研究注入了新的活力。主要对卷积神经网络在油井抽油机示功图自动识别中的应用进行研究,提出一种改进的Alexnet模型,实现了示功图的自动识别,并与目前常用的神经网络模型进行了比较。实验表明,改进的Alexnet模型在保证识别准确率高的同时有效降低了训练学习时间,很好地达到了实际应用要求。
Abstract
Nowadays, rod-pumping is still playing the most important role in oil production. Dynamometer card is the usual tool to detect, prevent and solve problems in oil production. Using AI method to analyze dynamometer card and detect malfunction of rod-pumping unit has been the focus of research. Deep learning has injected new energy into the study of the identification and interpretation of the dynamometer card. This paper mainly researched using convolutional neural networks in dynamometer card classification. We proposed an improved Alexnet model, and realized auto-classify of dynamometer card, then compared with classic convolutional neural network models. Experiments show that improved Alexnet model has high classification accuracy. Meanwhile reduces training time and reaches the practical application requirements.
下载PDF全文