Retrieval Augmented (Knowledge Graph), and Large Language Model-Driven Design Structure Matrix (DSM) Generation of Cyber-Physical Systems
A new method uses GraphRAG and LLMs to automatically map components in systems like CubeSats, with all code publicly released.
Researchers H. Sinan Bank and Daniel R. Herber published a paper demonstrating how Large Language Models (LLMs) combined with Graph-based Retrieval-Augmented Generation (GraphRAG) can automatically generate Design Structure Matrices (DSMs). They tested the method on two complex cyber-physical systems: a power screwdriver and a CubeSat satellite. The AI successfully identified components and their relationships, automating a key engineering design task. All code is publicly available for validation and further development by domain experts.
Why It Matters
This could drastically speed up the design phase for engineers building complex hardware and software systems.