Neural network meta-models for FPSO motion estimation from environmental data.

• Ano de defesa: 2023
• Autor(a) principal: Cotrim, Lucas Pereira
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: eng
• Instituição de defesa: Biblioteca Digitais de Teses e Dissertações da USP
• Palavras-chave em Português: Aprendizado computacional; Artificial neural networks; Engenharia Naval e Oceânica; Floating offshore platforms; Hyperparameter optimization; Inteligência artificial; Neural architecture search.; Otimização estocástica; Redes neurais; Surrogate models
• Link de acesso: https://www.teses.usp.br/teses/disponiveis/3/3152/tde-02062023-080712/

The current design process of mooring systems for Floating Production, Storage and Offloading units (FPSOs) is highly dependent on the availability of the platforms mathematical model and accuracy of dynamic simulations, through which resulting time series motion is evaluated according to design constraints. This process can be time-consuming and present inaccurate results due to the mathematical models limitations and overall complexity of the vessels dynamics. We propose a Neural Simulator, a set of data-based surrogate models with environmental data as input, each specialized in the prediction of different motion statistics relevant to mooring system design: Maximum Roll, Platform Offset and Fairlead Displacements. The meta-models are trained by current, wind and wave data given in 3h periods at the Campos Basin (Brazil) from 2003 to 2010 and the associated dynamic response of a spread-moored FPSO obtained through time-domain simulations using the Dynasim software. Hyperparameter Optimization techniques are performed in order to obtain optimal Artificial Neural Network (ANN) models specialized in different platform drafts. Finally, the proposed models are shown to correctly capture platform dynamics, providing good results when compared to motion statistics obtained from Dynasim. We conclude that an ANN surrogate model can be trained directly on real metocean conditions and corresponding measured FPSO motion statistics to provide increased accuracy and reduced computational time over traditional methods based on dynamic simulation. Moreover, the proposed architecture can be integrated into an automated learning framework: The data-based surrogate models can be continuously fine-tuned and updated with newly measured data, resulting in improved accuracy over time.