Nonlinear reduced DNN models for state estimation

DOI10.4208/CICP.OA-2021-0217MaRDI QIDQ5095220zbMATH OpenOpenAlexFDO

Authors

Publication date 5 August 2022

Published in Communications in Computational Physics (Search for Journal in Brave)

Full work available at URL https://arxiv.org/abs/2110.08951

zbMATH Keywords

deep neural networksnetwork expansionreduced basesResNet structuressensor coordinatesstate estimation in model-compliant norms

Mathematics Subject Classification ID

Second-order elliptic equations (35J15) Numerical methods for ill-posed problems for boundary value problems involving PDEs (65N20) Numerical methods for inverse problems for boundary value problems involving PDEs (65N21) Artificial neural networks and deep learning (68T07)

Abstract: We propose in this paper a data driven state estimation scheme for generating nonlinear reduced models for parametric families of PDEs, directly providing data-to-state maps, represented in terms of Deep Neural Networks. A major constituent is a sensor-induced decomposition of a model-compliant Hilbert space warranting approximation in problem relevant metrics. It plays a similar role as in a Parametric Background Data Weak framework for state estimators based on Reduced Basis concepts. Extensive numerical tests shed light on several optimization strategies that are to improve robustness and performance of such estimators.

Recommendations

• Nonlinear reduced models for state and parameter estimation
• Leveraging reduced-order models for state estimation using deep learning
• Optimal reduced model algorithms for data-based state estimation
• Non-intrusive model reduction of large-scale, nonlinear dynamical systems using deep learning
• Model reduction and neural networks for parametric PDEs

Cites work

• A parameterized-background data-weak approach to variational data assimilation: formulation, analysis, and application to acoustics
• Accuracy controlled data assimilation for parabolic problems
• Adaptive subgradient methods for online learning and stochastic optimization
• Approximation of high-dimensional parametric PDEs
• Convergence rates for greedy algorithms in reduced basis methods
• Data Assimilation in Reduced Modeling
• Data driven approximation of parametrized PDEs by reduced basis and neural networks
• Double greedy algorithms: Reduced basis methods for transport dominated problems
• Efficient online and batch learning using forward backward splitting
• Fast and Exact Simulation of Stationary Gaussian Processes through Circulant Embedding of the Covariance Matrix
• Generalized multiscale finite element method for the steady state linear Boltzmann equation
• scientific article; zbMATH DE number 1011659 (Why is no real title available?)
• Model reduction and neural networks for parametric PDEs
• Nonlinear reduced models for state and parameter estimation
• Optimal reduced model algorithms for data-based state estimation
• Reduced basis approximation and a posteriori error estimation for affinely parametrized elliptic coercive partial differential equations. Application to transport and continuum mechanics.
• Reduced Basis Greedy Selection Using Random Training Sets
• Space-time GMsFEM for transport equations
• Spatial Process Simulation

Cited in

(9)

• A learning-based projection method for model order reduction of transport problems
• State estimation with limited sensors -- a deep learning based approach
• Optimal reduced model algorithms for data-based state estimation
• Leveraging reduced-order models for state estimation using deep learning
• Nonlinear reduced models for state and parameter estimation
• State Estimation—The Role of Reduced Models
• Introduction: Wolfgang Dahmen's mathematical work (as of 2009)
• Expansive natural neural gradient flows for energy minimization
• 3D nonparametric neural identification

This page was built for publication: Nonlinear reduced DNN models for state estimation

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q5095220)