Year: 2024
Type: dissertation
Abstract: This thesis explores the integration of artifcial intelligence agents and large language models (LLMs) with functional mock-up units (FMUs) for human-centric decision support. The primary research question addresses how LLM agents can be efectively combined with FMUs to simplify the analysis of complex physical models and simulation results, making model-based decision support accessible for a broader group of users with the required domain knowledge. Although there is research on the general capabilities of LLMs in task solving, the specifc application of these models for managing large amount of parameters for tools like FMUs remains underexplored, marking a gap in both the deployment of LLMs in specialized scenarios and their adaptation to handle large sets of parameters. This investigation evaluates and explores practical strategies for implementing LLMs as assistants in simulation environments, where models in the form of FMUs can be integrated and analyzed using natural language. The research involved developing and testing a prototype interface that facilitates the interaction between LLMs and FMUs, complemented by case studies and a user survey to assess the system’s efectiveness. The fndings indicate that leveraging LLMs in conjunction with FMUs can streamline the process of setting up, executing, and interpreting simulations.
Author: Mattsson, Arvid
Language: English
Subfield: Modeling and Simulation
Field: Mathematics
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W7066706768