| 引用本文: | 李东泽,齐咏生,刘利强,马然,李永亭,刘思哲.基于双流生成对抗网络数据增强的风电机组智能故障诊断[J].电力自动化设备,2024,44(11):94-102. |
| LI Dongze,QI Yongsheng,LIU Liqiang,MA Ran,LI Yongting,LIU Sizhe.Intelligent fault diagnosis of wind turbine unit based on dual-stream generative adversarial network data augmentation[J].Electric Power Automation Equipment,2024,44(11):94-102. |
|
| |
|
|
| 本文已被：浏览 7102次 下载 1151次 |
码上扫一扫！ |
|
|
| 基于双流生成对抗网络数据增强的风电机组智能故障诊断 |
|
李东泽1,2,3, 齐咏生1,2,3, 刘利强1,2,3, 马然1,2,3, 李永亭1,2,3, 刘思哲1,2,3
|
|
1.内蒙古工业大学 电力学院，内蒙古自治区 呼和浩特 010080;2.大规模储能技术教育部工程研究中心，内蒙古自治区 呼和浩特 010080;3.内蒙古自治区高等学校智慧能源技术与装备工程研究中心，内蒙古自治区 呼和浩特 010080
|
|
| 摘要: |
| 针对实际工况下风电机组故障数据难以获取，现有数据增强方法对1维数据特征提取效果不佳的问题，提出一种基于双流生成对抗网络(DSGAN)的小样本智能故障诊断方法。设计了一种新的双流网络，通过深度特征提取流与时间特征提取流对风电机组故障数据进行深度与时间双特征提取。提出一种全局特征引导的自适应加权融合(GFG-AWF)模块对提取的双特征进行融合，并通过引入对抗生成思想，设计DSGAN完成小样本数据的增强。构建基于增强数据集辅助的双流诊断网络实现故障分类识别。利用轴承试验台数据与实际风电机组运行数据对所提方法进行了验证，最终诊断准确率达到98 %，表明所提方法可以有效解决小样本的故障诊断问题。 |
| 关键词: 风电机组 故障诊断 生成对抗网络 小样本 数据增强 |
| DOI：10.16081/j.epae.202407019 |
| 分类号:TM315 |
| 基金项目:国家自然科学基金资助项目(62363029,62241309)；内蒙古自然科学基金资助项目(2022MS06018,2021MS06018)；内蒙古自治区科技重大专项(2021ZD0040) |
|
| Intelligent fault diagnosis of wind turbine unit based on dual-stream generative adversarial network data augmentation |
|
LI Dongze1,2,3, QI Yongsheng1,2,3, LIU Liqiang1,2,3, MA Ran1,2,3, LI Yongting1,2,3, LIU Sizhe1,2,3
|
|
1.School of Electric Power, Inner Mongolia University of Technology, Hohhot 010080, China;2.Engineering Research Center of Large Energy Storage Technology, Ministry of Education, Hohhot 010080, China;3.Center for Intelligent Energy Technology and Equipment Engineering, Hohhot 010080, China
|
| Abstract: |
| In order to solve the difficulty in acquiring fault data of wind turbine under practical operating conditions and the poor effect of existing data augmentation methods on feature extraction of one-dimensional data, a small sample intelligent fault diagnosis method based on dual-stream generative adversarial network(DSGAN) is proposed. A new dual-stream network is designed to perform deep and temporal feature extraction on fault data of wind turbine through a deep feature extraction stream and a temporal feature extraction stream. A global feature guided adaptive weighted fusion(GFG-AWF) module is proposed to fuse the extracted dual features. By introducing adversarial generation principles, DSGAN is designed to enhance small-sample data. A dual-stream diagnostic network, assisted by the enhanced dataset, is constructed to achieve fault classification recognition. The proposed method is verified by bearing test-rig data and real wind turbine unit operational data, achieving a diagnostic accuracy of 98 %,which demonstrates that the proposed method can effectively solve the issue of fault diagnosis in small-sample scenarios. |
| Key words: wind turbine unit fault diagnosis generative adversarial network small sample data augmentation |
|
|
|
|
