Development of a FULly InTellIgent Maintenance FrAmework for PrognosTic Health ManagemEnt of Floating Offshore Wind Turbines (ULTIMATE)

This project aims to develop a fully intelligent solutions to the challenges of prognostic health management of floating offshore wind turbines (FOWT).

The research will develop a Physics Informed Deep Neural Network with Uncertainty Quantification (PIDNN-UQ) for real-time diagnosis and prognosis by combining smart and high precision dataset to address big data problem.

Numerical simulations of FOWT in coupled multi-physical fields will be conducted to investigate fatigue behaviours and mechanisms.

Smart (data-centric) databases of fatigue mechanisms for accurate modelling and analysis of FOWT will be devloped to facilitate realt-time diagnosis and prognosis.

The study will design and implement multi-tasking PIDNN-UQ models with physics-informed capability to improved model's knowledge and uncertainty quantification. This will enable the model to diagnose, quantify and predict the remaining useful lifetime of FOWTs.

Experimental and field data from Hywind 5 x 6MW FOWTs (Floating wind farm) and open source data from fixed bottom wind turbines (RAVE) would be used to validate and examine the performance of the ULTIMATE.

Outcomes of the research will contribute to advances in predictive maintenance, understanding of operation and performance of FOWT in real time.

This project will also contribute to knowledge in machine learning application to offshore engineering and renewable energy systems.

This will enhance curriculum development in O&M, structural integrity, data science and applied mathematics. engineering, mathematical theories of intelligent operation and maintenance of mechanical systems.

The project benefits the industry by developing intelligent maintenance methodologies based on PHM methods that delivers optimal FOWT operation with minimal human interface and improved safety and reliability.