Eggplant Team 2023
Virtual Wave Buoy: Using ML to Predict Wave Heights for Offshore Wind Energy Applications
The “Virtual Wave Buoy” project is a collaborative research initiative between Tufts Department of Civil and Environmental Engineering, Department of Electrical and Computer Engineering and industry partner Ørsted. The project aims to develop a machine learning predictive model called “Virtual Wave Buoy” to ensure safe and cost-effective offshore wind engineering operations. The Virtual Wave Buoy model learns the historical relationship between the sea state at an existing wave buoy/radar position and a network of public buoys to reproduce the data in case of equipment failure or removal. The project involves collecting and storing more buoy data from proprietary data sources from Ørsted and public, open-source datasets. The team has explored two sites off the east coast, the Gulf of Maine and the Martha’s Vineyard area, as well as several sites off the east and west coast of England, to train the model on various sea states. The model’s reliability will be improved and updated by implementing a transfer learning strategy between sites. The proof-of-concept code is written in Python and SQL has been developed to accommodate error handling, new sites, and comparison of several ML models. The project was supervised by Dr. Eleonora Tronci, a post-doc researcher at Tufts, and Ørsted engineers.
Related Tech Notes
- The Rise and Applications of Machine Learning on Wave Prediction by Seixas Aldrich
Project Poster
