Presentations

Abstract

Submarine power cables are critical infrastructure for offshore wind energy, and their reliability is essential for stable system operation. While typically buried to reduce damage risk, structural constraints around fixed-bottom wind turbines can prevent full burial. Inspections using Remotely Operated Vehicles (ROVs) are costly, limiting continuous and large-scale monitoring. Focusing on the communication optical fiber cables embedded within submarine power cables, we propose a cost-effective continuous monitoring method. Distributed Acoustic Sensing (DAS) is an emerging technique that measures dynamic strain and vibration along an optical fiber. The vibration response of submarine power cables depends on wave parameters and burial conditions. Spatio-temporal signals from multiple sections with varying exposure lengths derive a regression-based representation of cable exposure. Low-dimensional signal features extracted from this process, which can capture features related explicitly to exposure length and are robust to variations in measurement locations and wave conditions, are modeled as a normative reference for detecting exposure-length variations. To evaluate the performance, a power cable with embedded optical fibers was laid in a large tank equipped with a wave generator. Spatio-temporal signals were collected with exposed lengths ranging from 2 to 10 meters. The signed distance of a one-class SVM, used as an anomaly indicator in our approach, showed a mean difference of 0.05 between normal (no extension) and abnormal (≥1 meter extension) cases, indicating detectable cable extensions. The correlation coefficient of 0.71 between the exposure length estimated by the trained regression model and the actual length indicates that the low-dimensional signal features representing differences in exposure length were successfully extracted. The proposed method is expected to improve low-dimensional signal feature extraction with increasing data diversity. It will be further evaluated in real-world experiments to assess its effectiveness in anomaly detection and exposure-length estimation for submarine power cable monitoring.