计算机与数字工程2025，Vol.53Issue(2)：558-563,6.DOI:10.3969/j.issn.1672-9722.2025.02.045
基于迁移学习和微调的抽油机故障识别研究
Research on Fault Identification of Pumping Unit Based on Transfer Learning and Fine Tuning
摘要
Abstract
Aiming at the problems that the current fault detection methods on the well of pumping units mainly fail to achieve the real-time effect through manual field detection,and there are few pumping unit data available in the market,this paper propos-es a three-dimensional residual network model based on migration learning and fine tuning,fine tuning and retraining the high-lev-el convolution of the model through a small amount of data.Finally,the algorithm is optimized and implemented through the pytoch deep learning framework.The experimental results show that after 200 iterations,the different models tend to converge,and the pre-cision of the model test results after fine tuning is 20%higher than that of ordinary transfer learning.关键词
抽油机/深度学习/残差网络/迁移学习/故障识别/微调策略Key words
pumping unit/deep learning/residual network/transfer learning/fault identification/fine tuning strategy分类
信息技术与安全科学引用本文复制引用
李建平,吴江..基于迁移学习和微调的抽油机故障识别研究[J].计算机与数字工程,2025,53(2):558-563,6.基金项目
国家自然科学基金重点项目(编号:61933007)资助. （编号:61933007）
