MBD-NODE: Physics-informed data-driven modeling and simulation of constrained multibody systems

TL;DR

This paper introduces MBD-NODE, a physics-informed neural ODE framework that models multibody system dynamics by integrating physical constraints and prior knowledge, enabling more accurate and physically consistent data-driven simulations.

Contribution

It presents a novel neural ODE-based approach that incorporates physical constraints into multibody system modeling, improving upon existing neural network methods.

Findings

Successfully integrates physical constraints into neural ODE modeling.

Demonstrates improved accuracy over traditional neural network approaches.

Provides open-source code and data for reproducibility.

Abstract

We describe a framework that can integrate prior physical information, e.g., the presence of kinematic constraints, to support data-driven simulation in multi-body dynamics. Unlike other approaches, e.g., Fully-connected Neural Network (FCNN) or Recurrent Neural Network (RNN)-based methods that are used to model the system states directly, the proposed approach embraces a Neural Ordinary Differential Equation (NODE) paradigm that models the derivatives of the system states. A central part of the proposed methodology is its capacity to learn the multibody system dynamics from prior physical knowledge and constraints combined with data inputs. This learning process is facilitated by a constrained optimization approach, which ensures that physical laws and system constraints are accounted for in the simulation process. The models, data, and code for this work are publicly available as open source at https://github.com/uwsbel/sbel-reproducibility/tree/master/2024/MNODE-code.