摘要
油井功图是油井工作状态分析和故障诊断的重要依据,深度学习为油井功图的识别提供了有效手段,针对合理的深度神经网络架构的选择问题,构建了一个用于油井故障诊断的大型功图数据集,提出一种基于改进GoogLeNet网络结构的油井故障识别方法,并对深度神经网络的结构、激活函数、归一化层、训练方法、学习率等重要参数对识别精度和训练时间的影响进行了详细的分析.实验表明,相对于广泛使用的LeNet、ResNet和基本GoogLeNet等网络模型,提出的改进GoogLeNet网络模型有着更高的准确率;同时相对于基本GoogLeNet网络模型,所提模型的运行时间得到了有效的降低.
Oil well power diagram is very important for oil well working state analysis and fault diagnosis.Deep learning provides an effective means for oil well power diagram analysis,however,the optimal deep neural network architecture is difficult to determine.To solve this issue,first,a large power diagram data set for oil well fault diagnosis is constructed,second,an improved GoogLeNet structure is proposed for oil well fault identification.Meanwhile,the influence of core factors of a deep neural network,such as the activation function,the normalization layer,the training method,and the learning rate are analyzed.Experimental results show that the improved GoogLeNet network model has higher accuracy than the widely used network models such as LeNet,ResNet,and basic GoogLeNet.Compared with the basic GoogLeNet network model,the running time of the proposed model is effectively reduced.
作者
宋纯贺
李泽熙
于洪霞
刘意杨
冯铁英
张雪健
SONG Chunhe;LI Zexi;YU Hongxia;LIU Yiyang;FENG Tieying;ZHANG Xuejian(State Key Laboratory of Robotics,Shenyang Institute of Automation,Chinese Academy of Sciences,Shenyang 110016,China;Key Laboratory of Networked Control Systems,Chinese Academy of Sciences,Shenyang 110016,China;Institutes for Robotics and Intelligent Manufacturing,Chinese Academy of Sciences,Shenyang 110169,China;Shenyang Institute of Automation,Chinese Academy of Sciences,Shenyang 110016,China;School of Artificial Intelligence,Shenyang University of Technology,Shenyang 110870,China;Hangzhou XIOLIFT Co.Ltd.,Hangzhou 311199,China)
出处
《江苏科技大学学报(自然科学版)》
CAS
北大核心
2021年第2期52-58,共7页
Journal of Jiangsu University of Science and Technology:Natural Science Edition
基金
国家重点研发计划项目(2018YFB1700200)
国家自然科学基金资助项目(U1908212,61773368)
兴辽英才项目(XLYC1907057)。