Identification of Exposed Submarine Cable Status based on Optimized VMD-GAF-MCNN for Multi Sensor Fusion
【Objective】The multi-beam (side scan) sonar method is commonly employed to scan and measure the erosion of pile foundations and the landing sections of submarine cables. The degree of erosion is assessed based on the scanned images; however, this method lacks real-time capability, and relevant data...
Saved in:
| Main Authors: | , , , , , |
|---|---|
| Format: | Article |
| Language: | zho |
| Published: |
《光通信研究》编辑部
2025-06-01
|
| Series: | Guangtongxin yanjiu |
| Subjects: | |
| Online Access: | http://www.gtxyj.com.cn/thesisDetails#10.13756/j.gtxyj.2025.240060 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | 【Objective】The multi-beam (side scan) sonar method is commonly employed to scan and measure the erosion of pile foundations and the landing sections of submarine cables. The degree of erosion is assessed based on the scanned images; however, this method lacks real-time capability, and relevant data are only obtained during maintenance operations. Currently, temperature analysis and vibration analysis are frequently used for monitoring the erosion of submarine cables. However, the distributed optical fiber sensing signals collected on-site exhibit characteristics of non-stationarity, non-linearity, and susceptibility to noise interference. Additionally, there is an issue of incomplete feature information extracted by a single sensor. To address these challenges, a multi-sensor fusion method for recognizing the bare state of submarine cables is proposed. This method leverages optimized Variational Mode Decomposition (VMD), Gramian Angular Field (GAF), and Multi-scale Convolutional Neural Networks (MCNN) to enhance the accuracy of recognizing the bare state of submarine cables in the landing section.【Methods】The optimized VMD-GAF-MCNN method proposed in this paper integrates signal processing and deep learning techniques. Firstly, an optimized VMD method based on Hilbert Transform (HT) is introduced to calculate the maximum envelope kurtosis. This optimized VMD is used to decompose the distributed optical fiber sensing signal, and the correlation coefficient method is employed to screen the Intrinsic Mode Functions (IMFs). The IMF component with the highest correlation coefficient to the original signal is extracted. Secondly, a two-dimensional image encoding method using GAF is proposed. The selected IMF components from the optical fiber temperature and vibration signals, collected by Distributed Temperature Sensing (DTS) and Distributed Acoustic Sensing (DAS), are encoded into two-dimensional images using GAF. Finally, a MCNN structure is designed, with the training and test sets randomly divided. The training set is used to train the network, while the test set verifies the network's effectiveness in identifying the exposed state of the submarine cable.【Results】The verification was conducted using on-site collected temperature and vibration data from submarine cable fibers. The test accuracy achieved was 99.85%, which is 0.95% and 0.85% higher than that obtained using a single sensor, respectively.【Conclusion】The method proposed in this paper is capable of accurately identifying the exposed state of the submarine cable in the landing section, thereby providing robust technical support for developing operation and maintenance strategies for submarine cables. |
|---|---|
| ISSN: | 1005-8788 |