Deep residual shrinkage adaptive network-based cloud-edge-end collaborative fault diagnosis method for propulsion shafting system

工况号	转速/(r∙min⁻¹)	故障类型	样本数
1	200	正常/不对中/不平衡	100/100/100
2	250	正常/不对中/不平衡	100/100/100
3	300	正常/不对中/不平衡	100/100/100
4	350	正常/不对中/不平衡	100/100/100
5	400	正常/不对中/不平衡	100/100/100

Table 2. Setting the training parameter of RSN model
View table
View in Article
Table 2. Setting the training parameter of RSN model
参数数值
学习率 0.001
训练批次 5
迭代次数 50
损失函数交叉熵损失
优化器 Adam

Table 3. Average accuracy of ten diagnostic tests for unupdated and updated cloud and edge models

View table

View in Article

Table 3. Average accuracy of ten diagnostic tests for unupdated and updated cloud and edge models

模型	测试准确率
模型	工况1	工况2	工况3	工况4	工况5
初始未更新模型	0.931 1 ± 0.028 1	0.822 2 ± 0.011 3	0.926 7 ± 0.012 5	0.866 7 ± 0.016 3	0.844 4 ± 0.020 6
更新后云端模型	0.957 0 ± 0.020 6	0.962 0 ± 0.027 3	0.971 5 ± 0.012 8	0.931 5 ± 0.009 1	0.896 3 ± 0.023 1
更新后边缘模型	0.957 0 ± 0.020 6	0.962 0 ± 0.027 3	0.971 5 ± 0.012 8	0.931 5 ± 0.009 1	0.896 3 ± 0.023 1

Table 4. Parameter settings of online updating model
View table
View in Article
Table 4. Parameter settings of online updating model
参数数值
模型与环境每次交互样本数 5
每次更新交互次数 8
目标网络更新学习率 0.001
迭代更新次数 50
异步更新间隔 10

Tools

Get Citation

Copy Citation Text

Kaiwei YU, Zirui LI, Chong CHEN, Jun WU, Haiping ZHU. Deep residual shrinkage adaptive network-based cloud-edge-end collaborative fault diagnosis method for propulsion shafting system[J]. Chinese Journal of Ship Research, 2025, 20(4): 213

Download Citation

EndNote(RIS)BibTex Plain Text

Set citation alerts for article

Save article for my favorites