Project Details
Projekt
Translating thermodynamic knowledge to computers
Applicants
Professorin Dr. Sophie Fellenz; Professor Dr.-Ing. Hans Hasse; Professorin Dr. Heike Leitte
Subject Area
Chemical and Thermal Process Engineering
Image and Language Processing, Computer Graphics and Visualisation, Human Computer Interaction, Ubiquitous and Wearable Computing
Image and Language Processing, Computer Graphics and Visualisation, Human Computer Interaction, Ubiquitous and Wearable Computing
Term
since 2021
Project identifier
Deutsche Forschungsgemeinschaft (DFG) - Project number 466468799
This project addresses one of the fundamental topics of artificial intelligence: the translation of human knowledge to computers. Our ambition is to achieve this for the vast and deep knowledge on thermodynamics. Thermodynamics is particularly suited for this endeavor as it is a field with a highly developed and well-structured theory; and a successful translation would be particularly rewarding, as thermodynamics belongs to both science and engineering, and has extremely wide applications. It can be assumed that the basic principles that guide the translation of thermodynamic knowledge to computers can also be applied to other fields of science and engineering. In the first two years of the project, we have established a route to tackle this challenge, and have developed a system that enables computers to solve thermodynamic problems. That system, called KnowTD, has two basic parts: the ontology and the reasoner. The ontology comprises the thermodynamic theory as well as knowledge on material properties and thermodynamic problems. The reasoner defines the problem at hand based on the user input, then extracts the applicable equations from the ontology and solves them. In the second funding period, we want to extend the scope of KnowTD to open systems, to other materials than ideal gases, and to multistep problems, including cyclic processes, as well as to problems with coupled systems. We have also tested the ability of large language models (LLMs) such as ChatGPT to solve thermodynamic problems. For simple problems, ChatGPT 4 does a good job, but it fails when the problems become more complex. We want to extend this work in the second funding period, with the goal to take the best of both worlds and combine the flexibility of LLMs with the knowledge-based approach of KnowTD.
DFG Programme
Priority Programmes
