Abstract:Physics-constrained neural networks are commonly employed to enhance prediction robustness compared to purely data-driven models, achieved through the inclusion of physical constraint losses during the model training process. However, one of the major challenges of physics-constrained neural networks consists of the training complexity especially for high-dimensional systems. In fact, conventional physics-constrained models rely on singular-fidelity data necessitating the assessment of physical constraints within high-dimensional fields, which introduces computational difficulties. Furthermore, due to the fixed input size of the neural networks, employing multi-fidelity training data can also be cumbersome. In this paper, we propose the Multi-Scale Physics-Constrained Neural Network (MSPCNN), which offers a novel methodology for incorporating data with different levels of fidelity into a unified latent space through a customised multi-fidelity autoencoder. Additionally, multiple decoders are concurrently trained to map latent representations of inputs into various fidelity physical spaces. As a result, during the training of predictive models, physical constraints can be evaluated within low-fidelity spaces, yielding a trade-off between training efficiency and accuracy. In addition, unlike conventional methods, MSPCNN also manages to employ multi-fidelity data to train the predictive model. We assess the performance of MSPCNN in two fluid dynamics problems, namely a two-dimensional Burgers' system and a shallow water system. Numerical results clearly demonstrate the enhancement of prediction accuracy and noise robustness when introducing physical constraints in low-fidelity fields. On the other hand, as expected, the training complexity can be significantly reduced by computing physical constraint loss in the low-fidelity field rather than the high-fidelity one.

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

Data-driven simulation for general-purpose multibody dynamics using Deep Neural Networks

A Study of Mechanism-Data Hybrid-Driven Method for Multibody System Via Physics-Informed Neural Network

Novel Approach to Multibody System Modeling: Cascading and Clustering

A Parallel Coupling Framework for DEM-MBD: Model Verification and Application

Physically Consistent Neural ODEs for Learning Multi-Physics Systems

Visual simulation software for multibody system dynamics based on MBDyn

Multibody Dynamics Modeling and Simulating Using Maple and Maplesim

Research on Dynamic Modeling of Multi-body Dynamics Simulation Based on Inheritance and Transformation

Data driven modeling of heavy-duty joint system for DEMO manipulators: An initial study from MPD joint simulation

Dynamic Modeling Method for Constrained System with Singular Mass Matrices

An Efficient Hybrid Method for Multibody Dynamics Simulation Based on Absolute Nodal Coordinate Formulation

Real-Time Interactive Modeling and Simulation in Biological Physics and Active Matter with Mechanica

A Comprehensive Dynamic Simulation Framework for Coupled Neuromusculoskeletal-Exoskeletal Systems

Multi-fidelity physics constrained neural networks for dynamical systems

Dynamics Simulation with Unilateral Constraints in Virtual Assembly

New method for multibody dynamics based on unknown constraint force

The bionic neural network for external simulation of human locomotor system

Low-dimensional Dynamical Models of Structures with Uncertain Boundaries Via a Hybrid Knowledge- and Data-Driven Approach

Physics-data hybrid dynamic model of a multi-axis manipulator for sensorless dexterous manipulation and high-performance motion planning