A review of artificial intelligence methodologies in marine renewable energy technologies

Artificial intelligence (AI) is increasingly applied to marine renewable energy, but reported performance gains remain difficult to compare because studies differ in task, data regime, baseline, metric, and validation setting. This review synthesizes AI research published between 2010 and 2025 across wave energy, tidal energy, offshore wind, and ocean thermal energy conversion (OTEC). Using a structured coding template, we compare studies by engineering task, model family, decision role, data provenance, comparator, evaluation metric, validation context, and deployment implication. A five-level evidence ladder distinguishes numerical, laboratory, short-campaign, pilot-scale, and long-duration operational evidence. The synthesis shows that AI readiness is strongly domain-dependent. Offshore wind has the most mature evidence, particularly in operational forecasting and O&M, owing to SCADA, condition-monitoring, inspection, alarm, and maintenance data. Wave and tidal applications show promise in short-horizon forecasting, constrained control, and surrogate-assisted design, but remain largely site-specific and validation-limited. OTEC is the least mature domain, with evidence concentrated in component-level prediction, supervisory support, and experimentally bounded optimization. Emerging physics-informed, graph-based, generative, and foundation-model approaches broaden the toolkit, but their engineering value remains validation-dependent. The review contributes an evidence-weighted framework for assessing algorithmic promise, constrained proof of concept, and engineering readiness.