Artificial intelligence in offshore wind geotechnical engineering: A lifecycle-oriented review

As offshore wind develops toward deeper waters and larger turbine capacities, geotechnical challenges are becoming increasingly complex, involving more variable seabed conditions, larger foundation dimensions, and stricter lifecycle reliability and cost-control requirements. Accordingly, artificial intelligence (AI), with strengths in data fusion, surrogate modeling, and monitoring-informed inference, has increasingly been explored as a complementary computational and analytical tool in offshore wind geotechnical engineering. Particular attention is given to spatially variable ground characterization, surrogate-based response and reliability assessment, and monitoring-informed evaluation of installation and service states. This paper presents a lifecycle-oriented review of AI applications, informed by systematically screened research articles published between 2001 and 2025 and selected project- and industry-level practices. The main AI application pathways are synthesized across site investigation, foundation design, installation, and operation and maintenance (O&M), and are related to stage-specific geotechnical problems rather than algorithm categories alone. The synthesis shows a differentiated maturity pattern across the lifecycle. Foundation design shows the highest research concentration, while project-level digital ground modeling in site investigation and design-stage surrogate analysis show the clearest pathways to engineering use. Closed-loop installation control and below-mudline state inference for O&M remain largely at the research or validation stage. Across these applications, several cross-cutting limitations remain, including sparse offshore data, insufficient physical constraints, limited cross-site generalizability, incomplete standardization, and limited engineering acceptance. Future directions are discussed, including physics-integrated models, high-quality data ecosystems supported by standardized benchmark datasets, certification-oriented validation, and reviewable agent-assisted workflows to support more reliable and cost-effective offshore wind development.