Physics-Informed Multi-LSTM Networks for Metamodeling of Nonlinear Structures

By Ruiyang Zhang, Yang Liu. 2020

465 citations recorded by the source corpus.

This paper introduces an innovative physics-informed deep learning framework for metamodeling of nonlinear structural systems with scarce data. The basic concept is to incorporate physics knowledge (e.g., laws of physics, scientific principles) into deep long short-term memory (LSTM) networks, which boosts the learning within a feasible solution space. The physics constraints are embedded in the loss function to enforce the model training which can accurately capture latent system nonlinearity even with very limited available training datasets. Specifically for dynamic structures, physical laws of equation of motion, state dependency and hysteretic constitutive relationship are considered to construct the physics loss. In particular, two physics-informed multi-LSTM network architectures are proposed for structural metamodeling. The satisfactory performance of the proposed framework is successfully demonstrated through two illustrative examples (e.g., nonlinear structures subjected to ground motion excitation). It turns out that the embedded physics can alleviate overfitting issues, reduce the need of big training datasets, and improve the robustness of the trained model for more reliable prediction. As a result, the physics-informed deep learning paradigm outperforms classical non-physics-guided data-driven neural networks.

Subfields: Civil and Structural Engineering, Mechanical Engineering, Statistical and Nonlinear Physics

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Civil and Structural Engineering

In Mechanical Engineering

In Statistical and Nonlinear Physics

Citations in the corpus

Physics informed deep learning for computational elastodynamics without labeled data (2020, background): “uations [10,17{21], solving stochastic PDEs [22], ows transport in porous media [23{25], cardiovascular systems [26{28], design of metamaterials [29{ 31], metamodeling of nonlinear structural systems [32,33], and discovery of physical laws [34{36], among others. To further improve the learning perfo”
Physics-informed graph neural Galerkin networks: A unified framework for solving PDE-governed forward and inverse problems (2021, methodology): “Zhang et al. [22] presented a physics-guided CNN for seismic response modeling and also explored the similar idea with a Recurrent Neural Network (RNN) for metamodeling of nonlinear structures [23].”
PhyGeoNet: Physics-informed geometry-adaptive convolutional neural networks for solving parameterized steady-state PDEs on irregular domain (2020, background): “Most recently, researchers started to blend physics with CNN-based learning, such as imposing physical constraints into CNNs [57–63], development of physics-informed CNN for surrogate modeling [3, 6, 15, 64, 65], and discovering underlying governing equations from observed data within CNN-based arch”
Machine learning and structural health monitoring overview with emerging technology and high-dimensional data source highlights (2021, unknown): “Zhang et al., 342 proposed a deep LSTM network with the similar physics model approach as before.”
Scientific Machine Learning Through Physics–Informed Neural Networks: Where we are and What’s Next (2022, methodology): “For instance, Zhang et al (2020) observed that when a variable’s measurement was missing, the PINN implementation was capable of accurately predicting that variable.”