Abstract:The control of nonlinear large-scale dynamical models such as the incompressible Navier-Stokes equations is a challenging task. The computational challenges in the controller design come from both the possibly large state space and the nonlinear dynamics. A general purpose approach certainly will resort to numerical linear algebra techniques which can handle large system sizes or to model order reduction. In this work we propose a two-folded model reduction approach tailored to nonlinear controller design for incompressible Navier-Stokes equations and similar PDE models that come with quadratic nonlinearities. Firstly, we approximate the nonlinear model within in the class of LPV systems with a very low dimension in the parametrization. Secondly, we reduce the system size to a moderate number of states. This way, standard robust LPV theory for nonlinear controller design becomes feasible. We illustrate the procedure and its potentials by numerical simulations.

What problem does this paper attempt to address?

The paper attempts to address the computational challenges faced in control design for nonlinear large-scale dynamic models, such as the incompressible Navier-Stokes equations. Specifically, these challenges mainly arise from the potentially very large state space and nonlinear dynamics. To tackle these challenges, the authors propose a two-stage model reduction method tailored for the nonlinear controller design of incompressible Navier-Stokes equations and other partial differential equation models with quadratic nonlinearity. ### Main Issues: 1. **Large State Space**: Nonlinear systems typically have a very large state space, making direct controller design extremely difficult. 2. **Nonlinear Dynamics**: Nonlinear dynamics increase the complexity of the system, making traditional linear control theory hard to apply directly. 3. **Computational Efficiency**: It is necessary to reduce the computational burden while maintaining control performance to make controller design feasible in practical applications. ### Solution: 1. **Low-Dimensional Linear Parameter-Varying (LPV) Approximation**: First, approximate the nonlinear model as a low-dimensional LPV system to fit existing LPV control theory. 2. **System Scale Reduction**: Second, reduce the number of system states to a moderate level through model reduction techniques so that standard robust LPV control theory can be applied to nonlinear controller design. ### Specific Steps: 1. **Low-Dimensional Encoding**: Replace the nonlinear source term with a low-dimensional LPV surrogate model using the Proper Orthogonal Decomposition (POD) method. 2. **Parameter Domain Optimization**: Select an appropriate parameter domain (such as polytope or optimized polytope) to improve the performance and computational efficiency of the controller. 3. **LPV Controller Design**: Design the LPV controller using the `hinfgs` function in MATLAB's Robust Control Toolbox and evaluate its performance. ### Experimental Validation: The authors demonstrate the potential and feasibility of the method through numerical simulations and point out some limitations of existing methods, particularly the computational cost issue when dealing with large-scale Linear Matrix Inequalities (LMIs). ### Conclusion and Outlook: The method enables the application of existing LPV control theory and algorithms to the controller design of nonlinear systems through a two-stage model reduction. Future work will focus on improving the interpolation methods of controllers within the polytope and addressing the problem of large-scale Linear Matrix Inequalities.

Low-order Linear Parameter Varying Approximations for Nonlinear Controller Design for Flows

A Constructive Procedure Based on Dynamic Scaling for Adaptive Control of Nonlinearly Parameterized Nonlinear Systems

Low-complexity linear parameter-varying approximations of incompressible Navier-Stokes equations for truncated state-dependent Riccati feedback

Sequential Linear Quadratic Control of Bilinear Parabolic PDEs Based on POD Model Reduction

Convolutional Neural Networks for Very Low-Dimensional LPV Approximations of Incompressible Navier-Stokes Equations

NONLINEAR FLOW CONTROL USING A LOW DIMENSIONAL GALERKIN MODEL UDC 532.517:681.5

Nonlinear Tracking and Rejection using Linear Parameter-Varying Control

Sum-of-Squares approach to feedback control of laminar wake flows

Reduced-order controllers for fluid mechanics applications

An optimisation-based domain-decomposition reduced order model for parameter-dependent non-stationary fluid dynamics problems

Reduced‐order modeling for parametrized time‐dependent Navier‐Stokes equations

Reduced-order control using low-rank dynamic mode decomposition

Solving optimal control problems governed by random Navier-Stokes equations using low-rank methods

Space-time POD-Galerkin approach for parametric flow control

Deep polytopic autoencoders for low-dimensional linear parameter-varying approximations and nonlinear feedback design

Neural networks in feedback for flow analysis and control

Nonlinear parametric models of viscoelastic fluid flows

Two-stage model reduction approaches for the efficient and certified solution of parametrized optimal control problems

Neural networks in feedback for flow analysis, sensor placement and control

Reduced-order Deep Learning for Flow Dynamics

Turbulence Control Based on Reduced-Order Models and Nonlinear Control Design