Description

  ↵

Problem

Data-driven Digital Twins require operational data, such as event logs, to extract models of real-world systems. However, when introducing or reconfiguring a factory, historical data may not be available, limiting the early development of Digital Twins.

Goal

The goal of this thesis is to investigate how engineering knowledge,, such as Asset Administration Shell (AAS) descriptions, technical specifications, and layout information, can be used to bootstrap the development of a Digital Twin when operational data is not yet available.

Different approaches may be explored, including the use of AAS information, Generative AI and Large Language Models (LLMs), to support the generation of an initial Digital Twin. The thesis will investigate and evaluate the selected approach. 

The outcome of the thesis is a methodology for bootstrapping Digital Twin development from engineering knowledge, together with a prototype demonstrating its application.

Required Skills and Knowledge

•  Basic knowledge of Digital Twins and simulation.

• Programming skills in Python.

• Basic knowledge of Large Language Models and Generative AI.

• Interest in AI-based system design and data generation.

• Analytical thinking.

This thesis will be conducted in collaboration with Nanyang Technological University (NTU) Singapore, providing the opportunity for a research stay in Singapore as part of the thesis work