Chinese Journal of Ship Research, Volume. 18, Issue 5, 260(2023)
Self-attention and subdomain adaptive adversarial network for bearing fault diagnosis under varying operation conditions
Figures & Tables(14)
![[in Chinese]](/Images/icon/loading.gif){kind=icon}
Table 1. Parameters of SASAAN
View tableView in ArticleTable 1. Parameters of SASAAN
模块 网络结构 输出尺寸 通道数×内核尺寸 特征提取器 输入 64×64×3 — CBAM 64×64×32 32×5×5 CBAM 30×30×50 50×5×5 CBAM 13×13×64 64×4×4 子域自适应模块 FC1 600 — FC2 200 — FC3 3 — 域判别器 FC1 600 — FC2 200 — FC3 2 —
Table 2. The Ottawa bearing dataset
View tableView in ArticleTable 2. The Ottawa bearing dataset
数据集 健康状态 工况 样本量 渥太华 轴承数据集 健康（H） 内圈故障（I） 外圈故障（O） 加速（A） 2 400 减速（B） 2 400 先加速再减速（C） 2 400 先减速再加速（D） 2 400
Table 3. Experimental parameters
View tableView in ArticleTable 3. Experimental parameters
参数 数值 子域自适应强度系数 0.5 初始学习率 0.000 1 样本批量 40 最大训练次数/次 120 训练样本量/个 2400 600 测试样本量/个 1800
Table 4. Average diagnostic accuracy of twelve one-to-one transfer tasks on the target domain test set
View tableView in ArticleTable 4. Average diagnostic accuracy of twelve one-to-one transfer tasks on the target domain test set
迁移任务 目标域测试 准确率/% 迁移任务 目标域测试 准确率/% A→B 98.17 ±0.67 75.22±1.89 C→A 98.25 ±0.66 72.38±1.06 A→C 97.94 ±1.21 76.31±2.33 C→B 98.89 ±0.45 74.33±1.31 A→D 98.78 ±0.34 75.85±2.18 C→D 98.72 ±1.19 75.16±3.04 B→A 99.11 ±0.32 76.77±1.95 D→A 99.06 ±0.56 78.47±1.79 B→C 98.06 ±0.98 77.59±0.83 D→B 96.22 ±1.98 73.94±2.77 B→D 98.22 ±1.03 75.64±2.19 D→C 98.06 ±0.79 76.05±1.62
Table 5. Comparison of diagnostic accuracy of different methods
View tableView in ArticleTable 5. Comparison of diagnostic accuracy of different methods
方法 多工况迁移任务诊断精度/% 平均精度/% A→BCD B→ACD C→ABD D→ABC DCNN 76.11±2.42 75.16±1.62 78.14±3.32 75.34±1.94 76.19 DDC 87.46±2.16 83.43±0.98 88.36±1.69 86.78±3.43 86.51 DAN 92.44±1.27 88.11±1.45 91.12±0.76 90.17±0.56 90.46 DANN 95.39±0.78 94.12±1.06 95.35±0.93 94.28±1.46 94.79 SAAN 97.05 ±0.77 96.22 ±0.81 96.78 ±1.43 95.53 ±1.54 96.40 SASAAN 98.78 ±0.54 98.39 ±0.63 98.89 ±1.23 98.04 ±0.88 98.53
Tools
Get Citation
Copy Citation Text
Chao WANG, Bo TIAN, Zirui LI, Xiaoqi WANG, Jun WU. Self-attention and subdomain adaptive adversarial network for bearing fault diagnosis under varying operation conditions[J]. Chinese Journal of Ship Research, 2023, 18(5): 260
Paper Information
Category: Marine Machinery, Electrical Equipment and Automation
Received: Sep. 23, 2022
Accepted: --
Published Online: Mar. 21, 2025
The Author Email:
© Copyright 2018-2021 | Chinese Laser Press.
All Rights Reserved 沪ICP备15018463号-20