Searcharxiv⌕ Search

arXiv subjects

Matthew M. Peet

Publications and source records attributed to Matthew M. Peet.

At least 19 recordsLinked to original sources

A New Parametric Uncertainty Approach to Systems with Time-Varying Delay and Feedback Stabilization

Delay Differential Equations (DDEs) with an uncertain time-varying state delay are often modeled as a feedback interconnection of a nominal time-invariant system with a time-varying uncertainty operator. However, typical choices of uncertainty operators do not admit tight characterizations of input-output properties. In this paper, we propose a new representation of a DDE with time-varying delay system as a feedback interconnection of a nominal fixed delay DDE with a time-varying multiplicative parametric uncertainty operator, where the parameter is bounded in $L_\infty$ norms -- yielding an operator norm bound on the uncertainty. We then provide a modified uncertainty representation which uses the $L_2$ norm of the parameter to bound the operator norm of the uncertainty. Finally, we show that by representing the nominal system using Partial Integral Equations (PIEs), these characterizations of the uncertainty operator can be used for stability analysis or stabilizing feedback controller synthesis. The effectiveness of the approach is illustrated with simple numerical examples.

eess.SY↗

State-Dependent Delay Modeling of Wind-Farm Wake Dynamics

Turbines convert wind to electrical power. In the wake of a turbine, wind speed is reduced in proportion to the amount of power extracted. In a wind farm, downstream turbines may be located in wake of an upstream turbine, requiring coordination to yield maximum power generation. This paper creates a state-dependent delay model for a two turbine wind farm. The upstream turbine creates a wake that diminishes the wind speed at the downstream turbine. Changing the upstream turbine operation will affect the downstream turbine after a delay that depends on the wind-propagation time between them. This delay depends on the upstream power extraction and it can destabilize the combined dynamics. Conventional models of this effect rely on nonlinear partial differential equations (PDEs). Control design and analysis with such models is difficult with established methods. We first show that the PDE model can be represented by an Abel-type state-dependent delay differential equation. The equation has an implicit solution. To obtain an explicit representation, we then introduce a simplified state-dependent delay model. Both models predict destabilizing interactions between the two turbines. Numerical examples illustrate delay-induced oscillations using the simplified model and verify their effectiveness in simulation.

eess.SY↗

Dual Representations and Proxy Characterization of $H_2$ Norm for Partial Integral Equations

Duality and dual representations are essential tools for systems theory and controller synthesis. For any finite-dimensional causal linear time-invariant (CLTI) system, there exists a dual CLTI system with identical $H_\infty$ and $H_2$ norm. These relations are found by simply taking the adjoint of the transfer function. For nonlinear, time-varying or infinite-dimensional systems, however, such a transfer function may not exist or be difficult to obtain. For duality in the $H_\infty$ norm, this obstacle is overcome by using the equivalent time-domain, induced $L_2$-norm characterization. For $H_2$ norms there is no such equivalent characterization and hence proxy norms such as induced energy to peak and impulse to energy norms are used. These proxy norms bound the $H_2$ norm in the CLTI case. However, neither of these $H_2$ proxy norms equal the same $H_2$ proxy of the canonical dual representation. In this paper, we overcome this by proposing a dual representation whose induced energy to peak proxy equals the impulse to energy proxy of the primal and vice-versa. This relation is developed in the context of Partial Integral Equation (PIE) representations. The result is then used to establish primal-dual operator inequalities for both proxy norms similar in form to the primal-dual LMIs used to characterize the $H_2$ norm.

math.OC↗

Stable Rational Approximation of PDE Transfer Functions with $H_\infty$ Error Bounds

Transfer functions of Partial Differential Equations (PDEs) are irrational and difficult to obtain. Unlike for rational transfer functions, only a few methods are available for the control and analysis of irrational transfer functions. Thus, for robust analysis and control of PDEs, we need to use a rational approximation of the transfer function, preferably with provable $H_\infty$ error bounds. In this paper, we extend Krylov subspace model reduction to infinite-dimensional systems modeled by Partial Integral Equations (PIEs) -- an equivalent representation of PDEs. First, we propose a state transformation of the PIE system that allows us to compute a controllability matrix. Then, to obtain a rational approximation, we project the PIE states onto the vector space spanned by the truncated controllability matrix. Next, we show that sufficient conditions for exponential stability and well-posedness of the PIE system guarantee the exponential stability of the reduced-order system and provide constructive finite-frequency $H_\infty$ error bounds. Finally, we provide four illustrative numerical examples to demonstrate the performance of the proposed method.

math.AP↗

An Exact Lyapunov Characterization of Regional Rate Performance for Rational Stability

Rational stability is a quantitative stability notion for nonlinear ordinary differential equations (ODEs), under which convergence is characterized by a rationally-in-time decaying bound on solutions. Although Lyapunov characterizations of rational stability have been proposed, no existing condition has been shown to exactly characterize rational rate performance. The paper resolves this issue by providing an exact converse Lyapunov characterization of regional rational stability with prescribed rate performance. To test the resulting converse Lyapunov conditions numerically, a tiered set of convex relaxations is proposed which can be enforced using Sum-of-Squares (SOS) programming. A conjectured conservatism scaling factor in rate performance is associated with each tier and validated numerically. The first proposed tier appears to be novel and is shown to outperform classical SOS-based approaches. Numerical examples are used to validate the results and compute nested regions of state-space on which prescribed levels of performance can be guaranteed.

eess.SY↗

A State-Space Representation of Coupled Linear Multivariate PDEs and Stability Analysis using SDP

Recently, it has been shown that stability analysis and control of coupled Partial Differential Equations (PDEs) in a single spatial variable can be more conveniently performed using the Partial Integral Equation (PIE) representation. This PIE offers an equivalent, state-space representation of the PDE on the Hilbert space $L_{2}$, and is parameterized by an algebra of Partial Integral (PI) operators, allowing e.g. stability to be analyzed by solving a linear operator inequality on PI operator variables. In this paper, we show how this PIE framework for univariate PDEs can be extended inductively to multivariate PDEs on a hyper-rectangle. Specifically, assuming the boundary conditions defining the domain of the PDE to be decoupled along distinct spatial directions, we propose a readily verifiable condition for existence of a bijection between the PDE domain and $L_{2}$. We derive an explicit expression for the map defining this bijection, and use this map to construct an equivalent PIE representation for a broad class of linear multivariate PDEs. Next, we embed the parameters defining this PIE representation in a class of multivariate PI operators, and prove that this class forms a $*$-algebra---allowing PI operator inequalities for stability analysis, estimation, and control of univariate PDEs to be similarly formulated for multivariate PDEs. Finally, we show how such an operator inequality for stability analysis of multivariate PDEs can be solved with semidefinite programming, using a positive matrix parameterization of positive semidefinite multivariate PI operators. This framework for representation and stability analysis of multivariate PDEs is incorporated in the PIETOOLS software, and applied to analyze stability of 2D heat, wave, and plate equations, obtaining accurate bounds on the rate of decay.

math.AP↗

Verifying Well-Posedness of Linear PDEs using Convex Optimization

Ensuring that a PDE model is well-posed is a necessary precursor to any form of analysis, control, or numerical simulation. Although the Lumer--Phillips theorem provides necessary and sufficient conditions for well-posedness of dissipative PDEs, these conditions must hold only on the domain of the PDE---a proper subspace of $L_{2}$---which can make them difficult to verify in practice. In this paper, we show how the Lumer--Phillips conditions for PDEs can be tested more conveniently using the equivalent Partial Integral Equation (PIE) representation. This representation introduces a fundamental state in the Hilbert space $L_{2}$ and provides a bijection between this state space and the PDE domain. Using this bijection, we reformulate the Lumer--Phillips conditions as operator inequalities on $L_{2}$. We show how these inequalities can be tested using convex optimization methods, establishing an upper bound on the exponential growth rate of solutions. We demonstrate the effectiveness of the proposed approach by verifying well-posedness for several classical examples of parabolic and hyperbolic PDEs.

math.AP↗

Picard Iteration for Parameter Estimation in Nonlinear Ordinary Differential Equations using Low-Quality Data

We consider the problem of using experimental time-series data for parameter estimation in nonlinear ordinary differential equations, focusing on the case where the data is noisy, sparse, irregularly sampled, includes multiple experiments, and does not directly measure the system state or its time derivative. To account for such low-quality data, we propose a new framework for gradient-based parameter estimation which uses the Picard operator to reformulate the problem as constrained optimization with infinite-dimensional variables and constraints. We then formulate the Karush-Kuhn-Tucker (KKT) conditions necessary for optimality and define a convergent sequence of approximations to these KKT conditions obtained by replacing the solution map by the n-th order Picard iterate. Then, for any element of this sequence, and by exploiting the contractive properties of the Picard operator, we propose a gradient-contractive algorithm which (under regularity and convexity assumptions) is guaranteed to converge to a solution of these approximated KKT conditions. Finally, the algorithms are then tested on a battery of models and a variety of datasets in order to demonstrate robustness and improvement over alternative approaches.

math.OC↗

Hierarchical Stability and Lyapunov Conditions for Linear PDEs

Unlike ordinary differential equations (ODEs), linear partial differential equations (PDEs) admit multiple non-equivalent notions of stability. This variety makes interpretation of Lyapunov stability results challenging. \blue{To simplify this interpretation, we propose a unified framework for hierarchical classification of notions of stability and Lyapunov conditions. To do this, for classically well-posed PDEs with admissible boundary conditions}, we define a fundamental state on $L_2$ corresponding to the minimal information needed to uniquely forward propagate the solution. Stability notions and Lyapunov functions are then defined in terms of this fundamental state. This gives rise to a hierarchy of stability notions, the weakest being fundamental state to PDE state stability. Other stability notions and Lyapunov conditions may then be interpreted relative to this weakest notion. Hierarchies are established for: Lyapunov, exponential and finite-energy stability. Sufficient Lyapunov conditions are defined in terms of operator inequalities. Illustrative examples and computational tools are provided.

math.OC↗

Static Output Feedback Stabilization of Linear Systems with Multiple Delays

This work proposes a new procedure for the stabilization of time-delay systems using Static Output Feedback (SOF) control. A previous convex optimization approach to SOF for Ordinary Differential Equations (ODEs) is extended to time-delay systems through the use of a proposed state-space representation. This approach is based on solving two convex optimization problems, which are extensions of Linear Matrix Inequalities (LMIs) to infinite-dimensional systems. The first problem is stabilization under state feedback control; the second problem takes advantage of the Projection Lemma, which is extended here from matrices to Partial Integral (PI) operators. Finally, the results are compared with other SOF solutions for systems with delay found in the literature, showing a significant reduction in conservatism.

math.OC↗

$H_2$-Optimal Estimation of Linear Delayed and PDE Systems

The $H_2$ norm is a commonly used performance metric in the design of estimators. However, $H_2$-optimal estimation of most PDEs is complicated by the lack of transfer function and state-space representations. To address this problem, we first re-characterize the $H_2$-norm in terms of a map from initial condition to output. We then leverage the Partial Integral Equation (PIE) state-space representation of systems of linear PDEs coupled with ODEs to recast this characterization of $H_2$-norm as a convex optimization problem defined in terms of Linear Partial Integral (LPI) inequalities. We then parameterize a class of PIE-based observers and solve the associated $H_2$-optimal estimation problem. The observer synthesis problem is then recast as an LPI, and the resulting observers are validated using numerical simulation.

math.OC↗

Parameterization of Seed Functions for Equivalent Representations of Time-Varying Delay Systems

Abel's classic transformation shows that any well-posed system with time-varying delay is equivalent to a parameter-varying system with fixed delay. The existence of such a parameter-varying constant delay representation then simplifies the problems of stability analysis and optimal control. Unfortunately, the method for construction of such transformations has been ad-hoc -- requiring an iterative time-stepping approach to constructing the transformation beginning with a seed function subject to boundary-value constraints. Moreover, a poor choice of seed function often results in a constant delay representation with large time-variations in system parameters -- obviating the benefits of such a representation. In this paper, we show how the set of all feasible seed functions can be parameterized using a basis for $L_2$. This parameterization is then used to search for seed functions for which the corresponding time-transformation results in smaller parameter variation. The parameterization of admissible seed functions is illustrated with numerical examples that contrast how well-chosen and poorly chosen seed functions affect the boundedness of a time transformation.

math.OC↗

Time-Transformation-Based Analysis of Systems with Periodic Delay via Perturbative Expansion

It is difficult to analyze the stability of systems with time-varying delays. One approach is to construct a time-transformation that converts the system into a form with a constant delay but with a time-varying scalar appearing in the system matrices. The stability of this transformed system can then be analyzed using methods to bound the effect of the time-varying scalar. One issue is that this transformation is non-unique and requires the solution of an Abel equation. A specific time-transformation typically must be computed numerically. We address this issue by computing an explicit, although approximate, time-transformation for systems where the delay has a constant plus small periodic term. We use a perturbative expansion to construct our explicit solutions. We provide a simple numerical example to illustrate the approach. We also demonstrate the use of this time-transformation to analyze stability of the system with this class of periodic delays.

eess.SY↗

Lyapunov Functions can Exactly Quantify Rate Performance of Nonlinear Differential Equations

Pointwise-in-time stability notions for Ordinary Differential Equations (ODEs) provide quantitative metrics for system performance by establishing bounds on the rate of decay of the system state in terms of initial condition -- allowing stability to be quantified by e.g. the maximum provable decay rate. Such bounds may be obtained by finding suitable Lyapunov functions using, e.g. Sum-of-Squares (SOS) optimization. While Lyapunov tests have been proposed for numerous pointwise-in-time stability notions, including exponential, rational, and finite-time stability, it is unclear whether these characterizations are able to provide accurate bounds on system performance. In this paper, we start by proposing a generalized notion of rate performance -- with exponential, rational, and finite-time decay rates being special cases. Then, for any such notion and rate, we associate a Lyapunov condition which is shown to be necessary and sufficient for a system to achieve that rate. Finally, we show how the proposed conditions can be enforced using SOS programming in the case of exponential, rational, and finite-time stability. Numerical examples in each case demonstrate that the corresponding SOS test can achieve tight bounds on the rate performance with accurate inner bounds on the associated regions of performance.

math.OC↗

Efficient Convex Algorithms for Universal Kernel Learning

The accuracy and complexity of machine learning algorithms based on kernel optimization are determined by the set of kernels over which they are able to optimize. An ideal set of kernels should: admit a linear parameterization (for tractability); be dense in the set of all kernels (for robustness); be universal (for accuracy). Recently, a framework was proposed for using positive matrices to parameterize a class of positive semi-separable kernels. Although this class can be shown to meet all three criteria, previous algorithms for optimization of such kernels were limited to classification and furthermore relied on computationally complex Semidefinite Programming (SDP) algorithms. In this paper, we pose the problem of learning semiseparable kernels as a minimax optimization problem and propose a SVD-QCQP primal-dual algorithm which dramatically reduces the computational complexity as compared with previous SDP-based approaches. Furthermore, we provide an efficient implementation of this algorithm for both classification and regression -- an implementation which enables us to solve problems with 100 features and up to 30,000 datums. Finally, when applied to benchmark data, the algorithm demonstrates the potential for significant improvement in accuracy over typical (but non-convex) approaches such as Neural Nets and Random Forest with similar or better computation time.

stat.ML↗

Model Predictive Bang-Bang Controller Synthesis via Approximate Value Functions

In this paper, we propose a novel method for addressing Optimal Control Problems (OCPs) with input-affine dynamics and cost functions. This approach adopts a Model Predictive Control (MPC) strategy, wherein a controller is synthesized to handle an approximated OCP within a finite time horizon. Upon reaching this horizon, the controller is re-calibrated to tackle another approximation of the OCP, with the approximation updated based on the final state and time information. To tackle each OCP instance, all non-polynomial terms are Taylor-expanded about the current time and state and the resulting Hamilton-Jacobi-Bellman (HJB) PDE is solved via Sum-of-Squares (SOS) programming, providing us with an approximate polynomial value function that can be used to synthesize a bang-bang controller.

math.OC↗

$H_{\infty}$-Optimal Estimator Synthesis for Coupled Linear 2D PDEs using Convex Optimization

Any suitably well-posed PDE in two spatial dimensions can be represented as a Partial Integral Equation (PIE) -- with system dynamics parameterized using Partial Integral (PI) operators. Furthermore, $L_2$-gain analysis of PDEs with a PIE representation can be posed as a linear operator inequality, which can be solved using convex optimization. In this paper, these results are used to derive a convex-optimization-based test for constructing an $H_{\infty}$-optimal estimator for 2D PDEs. In particular, we first use PIEs to represent an arbitrary well-posed 2D PDE where sensor measurements occur along some boundary of the domain. An associated Luenberger-type estimator is then parameterized using a PI operator $\mathcal{L}$ as the observer gain. Examining the error dynamics of this estimator, it is proven that an upper bound on the $H_{\infty}$-norm of these error dynamics can be minimized by solving a linear operator inequality on PI operator variables. Finally, an analytical formula is proposed for inversion of a class of 2D PI operators, which is then used to reconstruct the Luenberger gain $\mathcal{L}$. Results are implemented in the PIETOOLS software suite -- applying the methodology and simulating the resulting observer for an unstable 2D heat equation with boundary observations.

math.OC↗

Constructive Representation of Functions in $N$-Dimensional Sobolev Space

A new representation is proposed for functions in a Sobolev space with dominating mixed smoothness on an $N$-dimensional hyperrectangle. In particular, it is shown that these functions can be expressed in terms of their highest-order mixed derivative, as well as their lower-order derivatives evaluated along suitable boundaries of the domain. The proposed expansion is proven to be invertible, uniquely identifying any function in the Sobolev space with its derivatives and boundary values. Since these boundary values are either finite-dimensional, or exist in the space of square-integrable functions, this offers a bijective relation between the Sobolev space and $L_{2}$. Using this bijection, it is shown how approximation of functions in Sobolev space can be performed in the less restrictive space $L_{2}$, reconstructing such an approximation of the function from an $L_{2}$-optimal projection of its boundary values and highest-order derivative. This approximation method is presented using a basis of Legendre polynomials and a basis of step functions, and results using both bases are demonstrated to exhibit better convergence behavior than a direct projection approach for two numerical examples.

math.NA↗