Leveraging reduced-order models for state estimation using deep learning
From MaRDI portal
Publication:5113091
Abstract: State estimation is key to both analyzing physical mechanisms and enabling real-time control of fluid flows. A common estimation approach is to relate sensor measurements to a reduced state governed by a reduced-order model (ROM). (When desired, the full state can be recovered via the ROM). Current methods in this category nearly always use a linear model to relate the sensor data to the reduced state, which often leads to restrictions on sensor locations and has inherent limitations in representing the generally nonlinear relationship between the measurements and reduced state. We propose an alternative methodology where a neural network architecture is used to learn this nonlinear relationship. Neural network is a natural choice for this estimation problem, as a physical interpretation of the reduced state-sensor measurement relationship is rarely obvious. The proposed estimation framework is agnostic to the ROM employed, and can be incorporated into any choice of ROMs derived on a linear subspace (e.g., proper orthogonal decomposition) or a nonlinear manifold. The proposed approach is demonstrated on a two-dimensional model problem of separated flow around a flat plate, and is found to outperform common linear estimation alternatives.
Recommendations
- Nonlinear reduced DNN models for state estimation
- State estimation with limited sensors -- a deep learning based approach
- Deep learning for model order reduction of multibody systems to minimal coordinates
- Reduced-order modeling of deep neural networks
- Non-intrusive model reduction of large-scale, nonlinear dynamical systems using deep learning
- Reduced-order deep learning for flow dynamics. The interplay between deep learning and model reduction
- A deep learning approach to Reduced Order Modelling of parameter dependent partial differential equations
- Reduced-order state estimation of delayed recurrent neural networks
- A multi-stage deep learning based algorithm for multiscale model reduction
- A two-stage deep learning architecture for model reduction of parametric time-dependent problems
Cites work
- A fast immersed boundary method using a nullspace approach and multi-domain far-field boundary conditions
- Compressive sensing based machine learning strategy for characterizing the flow around a cylinder with limited pressure measurements
- Data-Driven Sparse Sensor Placement for Reconstruction: Demonstrating the Benefits of Exploiting Known Patterns
- Deep learning
- Dynamic mode decomposition of numerical and experimental data
- Linearly recurrent autoencoder networks for learning dynamics
- Machine learning for fluid mechanics
- Model reduction of dynamical systems on nonlinear manifolds using deep convolutional autoencoders
- Near-Optimal Signal Recovery From Random Projections: Universal Encoding Strategies?
- Nonlinear model order reduction based on local reduced-order bases
- Shallow neural networks for fluid flow reconstruction with limited sensors
- Sparse reduced-order modelling: sensor-based dynamics to full-state estimation
- Sparse sensor placement optimization for classification
Cited in
(10)- Unsteady flow prediction from sparse measurements by compressed sensing reduced order modeling
- Dynamics of a data-driven low-dimensional model of turbulent minimal Couette flow
- Inadequacy of linear methods for minimal sensor placement and feature selection in nonlinear systems: a new approach using secants
- Data-driven reduced-order modeling of spatiotemporal chaos with neural ordinary differential equations
- Nonlinear reduced DNN models for state estimation
- Gappy AE: a nonlinear approach for gappy data reconstruction using auto-encoder
- Data-driven robust state estimation for reduced-order models of 2D Boussinesq equations with parametric uncertainties
- Shallow neural networks for fluid flow reconstruction with limited sensors
- State estimation with limited sensors -- a deep learning based approach
- Interpreted machine learning in fluid dynamics: explaining relaminarisation events in wall-bounded shear flows
This page was built for publication: Leveraging reduced-order models for state estimation using deep learning
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q5113091)
