Latent-Lagrangian Neural Networks for Reduced Order Modeling of Non-autonomous Nonlinear Dynamical Systems
My rating
This paper presents a latent Lagrangian-based framework for reduced-order modeling of forced nonlinear dynamical systems. The method jointly learns latent coordinates that capture system dynamics and two neural networks representing latent kinetic and potential energies. It uses force supervision to avoid requiring an ODE solver during training.
Tags: