Data-Driven Fault Diagnosis of Mooring Systems in Wave Energy Converters

Ocean renewable energy, particularly wave energy, is experiencing rapid growth in recent years. There is significant interest in moving ocean renewable technologies to offshore regions, given the higher resources (e.g., wave and wind), large area for deployment, less turbulence, and less negative environmental impacts. The mooring systems are the key functional component to guarantee the long-term reliability of floating structures, which, however, are prone to different types of failures. In addition to being consistently exposed to dynamic loads from waves and currents, the mooring lines are also exposed to damages from corrosion, biofouling, and bottom segment displacement. Therefore, it is critical to monitor the condition of the mooring lines' reliability for timely health management and maintenance and to avoid catastrophic failures. This research aims to achieve this objective by developing a new fault diagnosis framework that combines the Autoregressive (AR) model with Convolutional Neural Networks (CNN) to classify fault types and severity under random sea conditions. Two main fault types are considered in this study, including corrosion and biofouling, which are reflected in the numerical model developed for the RM3 Wave Energy Converter (specifically, stiffness decrease and mass increase for the mooring lines). The dynamic responses (surge, heave, pitch motions, and mooring line tensions) generated from this model will be utilized in the proposed fault diagnosis framework, with the AR model extracting features from time- domain data and being used as inputs to CNN for classification. It is noted that this approach not only addresses the challenges posed by random phase shifts in ocean waves but also significantly reduces the computational demand, thereby streamlining the training process and improving the accuracy of fault detection. The simulation results indicate an accurate prediction of the fault type and severity under highly random sea conditions, which demonstrates the feasibility of the proposed method.