Abstract
Marine renewable energy (MRE) systems operate in harsh marine environments where long-term exposure to seawater leads to biofouling, resulting in increased surface roughness, hydrodynamic drag, added mass, structural loading, sensor degradation, and reduced energy production. Despite its significant operational and economic impact, biofouling management in MRE devices has traditionally relied on manual inspections and empirical growth models, which offer limited predictive capability. This review provides a structured, data-centric synthesis of recent advances in machine learning (ML) and bioinformatics approaches for biofouling modeling, performance prediction, and maintenance planning in offshore wind turbines, tidal turbines, and wave energy converters. The study systematically examines key fouling locations and associated engineering impacts, and analyzes the major data streams used for predictive modeling, including SCADA and condition-monitoring time series, metocean variables, inspection imagery, laboratory and field experiments, and environmental DNA (eDNA) sequencing outputs. We compare modeling strategies ranging from physics-based simulations to classical ML, deep learning, computer vision, and hybrid physics-informed frameworks, and discuss how biological indicators such as microbial community profiles and eDNA-derived taxa abundances can be integrated as predictive features. The review further outlines emerging digital twin architectures for fouling-aware performance forecasting and maintenance decision support. Finally, we identify key challenges including data scarcity, cross-site generalization, validation practices, and uncertainty quantification, and propose future research directions toward integrated, proactive biofouling management systems in marine renewable energy infrastructure.