Abstract:Finite element models (FEM) of geometrically nonlinear structures are increasingly computationally expensive as physical systems become more complex. Nonlinear reduced order models (NLROM) are an efficient alternative to simulate the dynamic behavior of practical engineering structures. However, the computation complexity of ROM increases exponentially with the number of selected modes, which makes the calculation effort very expensive. This study proposes a fusion method that combines reduced-order and predictive modeling, referred to as the multi-fidelity surrogate-based nonlinear reduced-order model (MFS-NLROM). The MFS-NLROM approach defines the nonlinear solution as the high-fidelity response and the linear solution as the low-fidelity response, enabling faster calculation of nonlinear FEM. By incorporating a discrepancy function to represent the nonlinear force, the proposed method increases the complexity of the fitting function while reducing the number of nonlinear terms. This simplifies nonlinear problems through the predicted model of the nonlinear force. The MFS-NLROM method is applicable to static, dynamic, and nonlinear normal mode analyses. In dynamic analysis, error compensation is introduced through the fusion of the MFS-NLROM and Newmark methods. Additionally, the first derivative of the multi-fidelity surrogate is utilized to obtain nonlinear normal modes (NNMs) using the multi-harmonic balance (MHB) methods, which enables the evaluation of the accuracy of the proposed method. The feasibility and effectiveness of the proposed framework are verified using examples involving a flat beam and a wing cantilever beam, in which the Implicit Condensation and Expansion (ICE) model and MFS-NLROM are fully compared. The evaluation results show that MFS-NLROM has better computation efficiency under the premise of ensuring accuracy. In short, the proposed framework can leverage the advantages of computational accuracy and efficiency, which can serve as a means to promote the development of digital twins in industrial fields (dynamic system damage monitoring, safety assessment, and other aspects).

A nonintrusive nonlinear model reduction method for structural dynamical problems based on machine learning

Regression-based Identification and Order Reduction Method for Nonlinear Dynamic Structural Models

A local basis approximation approach for nonlinear parametric model order reduction

Non-intrusive model reduction of large-scale, nonlinear dynamical systems using deep learning

Lagrangian operator inference enhanced with structure-preserving machine learning for nonintrusive model reduction of mechanical systems

Model Order Reduction based on Runge-Kutta Neural Network

Low-dimensional Data-based Surrogate Model of a Continuum-mechanical Musculoskeletal System Based on Non-intrusive Model Order Reduction

Nonlinear Manifold Learning and Model Reduction for Transonic Flows

Nonintrusive Structural Dynamic Reduced Order Modeling for Large Deformations: Enhancements for Complex Structures

A Review of Model Order Reduction Methods for Large-Scale Structure Systems

Machine Learning Approach to Model Order Reduction of Nonlinear Systems via Autoencoder and LSTM Networks

Randomized low‐rank approximation methods for projection‐based model order reduction of large nonlinear dynamical problems

A Novel Geometric Nonlinear Reduced Order Modeling Method Using Multi-Fidelity Surrogate for Real-Time Structural Analysis

A Hybrid Reduced-Order Modeling Technique for Nonlinear Structural Dynamic Simulation

A state-of-the-art survey of model order reduction techniques for large-scale coupled dynamical systems

A manifold learning approach to nonlinear model order reduction of quasi-static problems in solid mechanics

A nonintrusive distributed reduced‐order modeling framework for nonlinear structural mechanics—Application to elastoviscoplastic computations

Projection-Based Model Reduction with Dynamically Transformed Modes

An Adaptive Substructure -Based Model Order Reduction Method for Nonlinear Seismic Analysis in OpenSees

Learning physics-based reduced-order models from data using nonlinear manifolds

Model order reduction of deep structured state-space models: A system-theoretic approach