GRU neural network-based method for box girder crack damage detection

Table 1. The material parameters of the box girder
View table
View in Article
Table 1. The material parameters of the box girder
参数数值
密度$ \rho $/(kg·m−3) 7.85×103
杨氏模量E/MPa 2.058×105
泊松比ν 0.3
$ \alpha $/s−1 14.82
$ \beta $/s 3.59×10−7

Table 2. The hyperparameters of the GRU model
View table
View in Article
Table 2. The hyperparameters of the GRU model
超参数数值
学习率 0.01
训练次数 50
批处理大小 32
输入层维度 22
隐藏层特征维度 32
循环层数 2
舍弃概率 0.5

Table 3. The prediction accuracy of crack location
View table
View in Article
Table 3. The prediction accuracy of crack location
方法预测精度/%
无噪声有噪声
WPT-MLP 94.63 76.45
GRU 100.00 97.14

Table 4. Confusion matrix of crack location prediction using GRU method with noise

View table

View in Article

Table 4. Confusion matrix of crack location prediction using GRU method with noise

实际标签	预测标签											总计	准确率/%
实际标签	0	1	2	3	4	5	6	7	8	9	10	总计	准确率/%
0	306	2	3	12	9	1	0	1	0	1	0	335	91.34
1	4	245	0	0	2	0	0	0	0	1	0	252	97.22
2	0	0	174	0	0	0	0	0	0	0	0	174	100.00
3	10	0	0	297	0	0	0	0	0	0	0	307	96.74
4	5	1	0	0	301	1	0	0	0	0	0	308	97.73
5	12	0	3	0	2	305	0	0	0	0	0	322	94.72
6	0	0	0	0	1	0	238	0	0	0	0	239	99.58
7	1	0	0	0	0	0	0	276	0	2	0	279	98.92
8	0	0	0	0	0	0	0	0	280	0	0	280	100.00
9	2	2	0	0	0	0	0	0	2	194	0	200	97.00
10	2	0	0	0	0	0	1	0	0	0	205	208	98.56
总计	342	250	180	309	315	307	239	277	282	198	205	2 904	97.14

Table 5. Confusion matrix of crack location prediction using WPT-MLP method with noise

View table

View in Article

Table 5. Confusion matrix of crack location prediction using WPT-MLP method with noise

实际标签	预测标签											总计	准确率%
实际标签	0	1	2	3	4	5	6	7	8	9	10	总计	准确率%
0	162	16	34	26	23	27	6	15	8	13	5	335	48.36
1	29	195	9	3	1	7	1	5	0	2	0	252	77.38
2	3	10	158	2	1	0	0	0	0	0	0	174	90.80
3	24	14	2	233	11	12	1	3	1	4	2	307	75.90
4	34	4	8	7	208	37	2	2	1	2	3	308	67.53
5	35	3	2	1	34	235	5	1	4	2	0	322	72.98
6	7	1	4	0	1	8	199	3	5	0	11	239	83.26
7	21	4	0	7	4	2	2	228	10	1	0	279	81.72
8	6	1	1	7	1	1	0	4	255	0	4	280	91.07
9	28	9	3	1	3	4	0	0	0	152	0	200	76.00
10	2	0	6	1	0	0	3	0	1	0	195	208	93.75
总计	351	257	227	288	287	333	219	261	285	176	220	2 904	76.45

Table 6. The prediction accuracy of crack length
View table
View in Article
Table 6. The prediction accuracy of crack length
方法预测精度/%
无噪声有噪声
WPT-MLP 74.04 66.77
GRU 96.38 88.67

Table 7. Confusion matrix of crack length prediction using GRU method with noise
View table
View in Article
Table 7. Confusion matrix of crack length prediction using GRU method with noise
实际标签预测标签总计准确率%
0 1 2 3
0 191 91 0 0 282 67.73
1 96 851 2 0 949 89.67
2 0 28 812 71 911 89.13
3 0 0 41 721 762 94.62
总计 287 970 855 792 2 904 88.67

Table 8. Confusion matrix of crack length prediction using WPT-MLP method with noise
View table
View in Article
Table 8. Confusion matrix of crack length prediction using WPT-MLP method with noise
实际标签预测标签总计准确率%
0 1 2 3
0 7 265 10 0 282 2.48
1 11 762 158 18 949 80.30
2 1 124 548 238 911 60.15
3 0 14 126 622 762 81.63
总计 19 1 165 842 878 2 904 66.77

Tools

Get Citation

Copy Citation Text

Xiedong LUO, Dongliang MA, Songlin ZHANG, Deyu WANG. GRU neural network-based method for box girder crack damage detection[J]. Chinese Journal of Ship Research, 2022, 17(4): 194

Download Citation

EndNote(RIS)BibTex Plain Text

Set citation alerts for article

Save article for my favorites

Paper Information

Category: Ship Structure and Fittings

Received: Jun. 12, 2021

Accepted: --

Published Online: Mar. 26, 2025

The Author Email:

DOI:10.19693/j.issn.1673-3185.02415

Topics

laser devices and laser physics

Lasers and Laser Optics

Laser physics

laser manufacturing

Instrumentation, Measurement and Metrology

Table 1. The material parameters of the box girder

Table 1. The material parameters of the box girder

Table 2. The hyperparameters of the GRU model

Table 2. The hyperparameters of the GRU model

Table 3. The prediction accuracy of crack location

Table 3. The prediction accuracy of crack location

Table 4. Confusion matrix of crack location prediction using GRU method with noise

Table 4. Confusion matrix of crack location prediction using GRU method with noise

Table 5. Confusion matrix of crack location prediction using WPT-MLP method with noise

Table 5. Confusion matrix of crack location prediction using WPT-MLP method with noise

Table 6. The prediction accuracy of crack length

Table 6. The prediction accuracy of crack length

Table 7. Confusion matrix of crack length prediction using GRU method with noise

Table 7. Confusion matrix of crack length prediction using GRU method with noise

Table 8. Confusion matrix of crack length prediction using WPT-MLP method with noise

Table 8. Confusion matrix of crack length prediction using WPT-MLP method with noise