Physics-informed machine learning applied for forecasting metocean conditions

• Grant number: 20/16746-5
• Support Opportunities: Scholarships in Brazil - Doctorate
• Effective date (Start): March 01, 2021
• Effective date (End): February 29, 2024
• Field of knowledge: Engineering - Electrical Engineering
• Acordo de Cooperação: IBM Brasil
• Principal Investigator: Eduardo Aoun Tannuri
• Grantee: Felipe Marino Moreno
• Host Institution: Escola Politécnica (EP). Universidade de São Paulo (USP). São Paulo , SP, Brazil
• Host Company: Universidade de São Paulo (USP). Centro de Inovação da USP (INOVA)
• Associated research grant: 19/07665-4 - Center for Artificial Intelligence, AP.PCPE

Abstract

Operations at the sea are heavily affected by the environmental conditions (currents, wave and winds) and for them to be safely planned an accurate forecast of the conditions is needed. Numerical simulations based on physical equations are often employed to forecast these those conditions, but their accuracy are limited by the accuracy and amount of known information about forcing and boundary conditions as well. One approach to improve the forecast accuracy is to combine the physical model with machine learning algorithms (this approach is also known as Physics-Informed Machine Learning - PILM), taking advantage of the pattern recognition power of Machine Learning while keeping the physical consistency of the physical model. This work aims to use PILM to improve metocean forecasting in relevant regions near the Brazilian Coast. (AU)