Executive Summary : | India has committed to reducing carbon emissions by 33-35% below 2005 levels by 2030. Wind energy is crucial to achieving this goal, and India aims to increase the share of wind energy in its energy mix to 40% by 2030. Proper maintenance of wind turbines is important for sustained energy production. Floating Offshore Wind Turbines (FOWT) are becoming popular due to issues with land acquisition, aesthetic pollution, and moderate energy yield. The proposal offers a non-intrusive continual targeted CA approach for component-wise health estimation for FOWTs that combines physics-based and data-driven approaches through a physics-informed learning approach. Three subsystems, namely: blades, tower, and mooring lines will be investigated in this attempt. Physics-informed neural networks (PINNs) can revolutionize the structural health monitoring of offshore wind turbines by combining low-fidelity models with sparse high-fidelity real data and especially when that can be employed at the component level while ignoring the rest of the structure. This study proposes the use of non-intrusive sensing and computer vision techniques based on machine learning to extract response information from high-speed video data. The proposed algorithm will address the SHM of FOWTs by estimating location-based health parameters alongside the network parameters, yielding probabilistic measures for the material/health parameters within a Bayesian learning environment. The algorithm will be validated against high-end numerical models and further investigated through noise sensitivity study, model dependency, and parametric analysis to investigate its real-life applicability. |