This paper proposes an Augmented Semantic Digital Twins (LSDTs) framework that leverages large-scale language models (LLMs) to address the challenge of integrating unstructured knowledge and build effective digital twins (DTs) for managing complex infrastructure systems. LSDTs leverage LLMs to extract planning knowledge from unstructured documents, such as environmental regulations and technical guidelines, and organize it into a formal ontology. This ontology forms a semantic layer that powers digital twins, virtual models of physical systems, enabling the simulation of realistic and compliant planning scenarios. Case studies of the Maryland offshore wind farm planning and Hurricane Sandy applications demonstrate the effectiveness of LSDTs, demonstrating their ability to support interpretable and compliant layout optimization, high-fidelity simulations, and enhanced adaptability of infrastructure planning. In conclusion, the combination of generative AI and digital twins demonstrates the potential to support complex, knowledge-based planning tasks.
