Laboratory for Dynamics of Machines and Structures 
Automated discovery of fundamental variables hidden in experimental data

To find the fundamental state variables of complex dynamical systems, an experienced engineer must study the dynamics of the system, which often requires analytical, numerical, and experimental approaches. The question is whether it is possible to find the state variables from the recording of the response of a dynamical system using an automatic algorithm based on a neural network.
The task requires an in-depth study of the existing algorithm in an article published in Nature in 2022. Followed by the development of a realistic dynamic system with multiple degrees of freedom to test the algorithm.

Project webpage

Contact: Janko Slavič