Partial Floquet Transformation and Model Order Reduction of Linear Time-Periodic Systems

Sam Bender, Christopher Beattie · 2025

Read Partial Floquet Transformation and Model Order Reduction of Linear Time-Periodic Systems by Sam Bender, Christopher Beattie (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2508-00221.txt.

Introduction

Sam Bender and Christopher Beattie work on shrinking huge models of systems whose behavior repeats in time, in a 2025 paper. Such time-periodic systems arise when a nonlinear system is linearized around a repeating motion, like a rotor or a power grid cycle. Engineers need small versions to test many inputs fast. The classic shrinking tool, the Floquet transform, is too costly to compute for big systems.

The pair define a partial version of that transform, built only on the most useful directions of the system's motion. They adapt a known pole-hunting algorithm to find those directions. From them they build time-varying bases that yield small models which still behave like the original. A worked example shows the recipe in action. The authors aim at fast simulation for large periodic systems in engineering.

Catalog record

CREATORSam Bender, Christopher Beattie
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE57 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2508-00221

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Partial Floquet Transformation and Model Order Reduction of Linear Time-Periodic Systems Sam Bender, Christopher Beattie Abstract Time-periodic dynamical systems occur commonly both in nature and as engineered systems. Large-scale linear time-periodic dynamical systems, for example, may arise through linearization of a nonlinear system about a given periodic solution (possibly as a consequence of a baseline periodic forcing) with subsequent spatial discretization. The potential need to simulate responses to a wide variety of input profiles (viewed as perturbations off a baseline periodic forcing) creates a potent incentive for effective model reduction strategies applicable to linear time-periodic ( ltp ) systems. Classical approaches that take into account the underlying time-periodic system structure often utilize the Floquet transform; however, computation of the Floquet transform is typically intractable for large order systems. In this paper, we develop the notion of a partial Floquet transformation connected to selected invariant subspaces of a time-varying differential operator associated with the ltp system. We modify and repurpose the Dominant Pole Algorithm of Rommes to identify effective invariant subspaces useful for model reduction. We discuss the construction of associated partial Floquet transformations and time-varying reduction bases with which to produce effective reduced-order ltp models and illustrate the process on a simple time-periodic system. keywords Model Order Reduction, Periodic Dynamical Systems, Dominant Pole Algorithm, Floquet Transform, Harmonic Transfer Function. 1 Introduction Time-periodic dynamical systems occur commonly, both in nature and as engineered systems, often as a consequence of periodic forcing due to rotation (e.g., the Earth’s rotation generates both tidal gravity forces and diurnal temperature gradients that cyclically drive atmospheric and ocean flows; gyroscopic forces can generate periodic forcing causing significant vibration and noise in vehicles). More broadly, periodic phenomena can occur through the emergence of a dynamic balance between inertial and various restoring forces. For example, a structure exposed to an otherwise steady wind- or current-flow can experience large oscillations caused by vortex shedding or flutter. This can be destructive (the Tacoma Narrows bridge failure is a famous example); but there can be positive effects as well (e.g., high-efficiency wind turbines may take advantage of these effects). Linear time-periodic ( ltp ) systems play a fundamental role in the analysis, simulation, and control of such phenomena even when the underlying models reflect fundamentally nonlinear dynamics, since by their character the periodic phenomena of interest emerge as components of a “center manifold” and must themselves be stable at least when subjected to small perturbations. Were this not the case, say for a natural system, oscillatory phenomena would not generally be observed, while for an engineered system, they would not generally be desired. Beyond this, computational strategies for extracting periodic solutions of nonlinear systems necessitate repeated solution of linear(ized) periodic systems, and this leads (naturally) to the the question of effective model order reduction strategies for such systems. See, e.g., Bittanti and Lovera 1996 ; Gazzola et al. 2022 ; Möllerstedt and Bernhardsson 2000 . In Section 2, we introduce the analytic setting that provides the framework for our work. We also define the partial Floquet transformation and connect it with the classical Floquet transformation. The framework developed in Section 2 identifies invariant subspaces of a time-varying differential operator associated to the ltp system dynamics with families of reduction bases and hence to classes of ltp -reduced-order models. In Section 3, we consider the problem of identifying “good” reduced-order models by identifying in turn “good” invariant subspaces. This is done by developing notions of pole dominance in an ltp setting. In Section 4, we then modify and repurpose the Dominant Pole Algorithm of Rommes Rommes 2007 to account for our notion of ltp - pole dominance and use it to identify effective ltp -reduced-order models. We illustrate its use on a simple example in Section 5 and provide concluding remarks in Section 6. 2 Problem Setting We consider time-periodic dynamical systems of the form: G:\left\{\begin{aligned} \dot{\mathbf{x}}(t)&=\mathbf{A}(t)\mathbf{x}(t)+\mathbf{b}(t)u(t)\\ y(t)&={\bf c}(t)^{*}\mathbf{x}(t)\end{aligned},\right. (1) where for each t\in\mathbb{R} , \mathbf{A}(t)\in\mathbb{C}^{n\times n} and \mathbf{b}(t),\mathbf{c}(t)\in\mathbb{C}^{n} . We further assume that \mathbf{A}(t) , \mathbf{b}(t) , and \mathbf{c}(t) are locally integrable and T -periodic (i.e., \mathbf{A}(t)=\mathbf{A}(t+T) , \mathbf{b}(t)=\mathbf{b}(t+T) and \mathbf{c}(t)=\mathbf{c}(t+T) t -almost everywhere for a fixed T>0 ), and so, the Carathéodory conditions for existence and uniqueness of solutions \mathbf{x}(t) hold. For any given order r\ll n , we seek a reduced-order time-periodic linear model , \displaystyle\dot{\mathbf{x}}_{r}(t) \displaystyle=\mathbf{A}_{r}(t)\mathbf{x}_{r}(t)+\mathbf{b}_{r}(t){u}(t), (2) \displaystyle{y}_{r}(t) \displaystyle=\mathbf{c}_{r}(t)^{*}\mathbf{x}_{r}(t) where \mathbf{A}_{r}(t)\in\mathbb{C}^{r\times r} and \mathbf{b}_{r}(t),\,\mathbf{c}_{r}(t)\in\mathbb{C}^{r} are obtained via projection so as to be also locally integrable and T -periodic, and furthermore defined in such a way so that {y}_{r}(t)\approx{y}(t) over a wide class of admissible inputs {u}(t) . 2.1 The Floquet Transform Given an ltp system as in ( 1 ), the Floquet transformation is defined via the monodromy matrix , i.e., the fundamental solution matrix for the homogeneous system evaluated at T (the system period); see e.g., Chicone 2006 . The fundamental solution matrix is an absolutely continuous matrix-valued function \Phi(t)\in\mathbb{C}^{n\times n} such that \Phi(0)=\mathbf{I} and \dot{\Phi}=\mathbf{A}(t)\Phi ; the monodromy matrix is then \Phi(T) . With a suitable choice of branch cut for the (complex) logarithm, one may define \mathbf{R}=\frac{1}{T}\log\Phi(T) from which it follows that \mathbf{P}(t)=\Phi(t)\exp(-t\mathbf{R}) is T -periodic. This leads to a time-periodic change of variable \mathbf{z}(t)=\mathbf{P}(t)^{-1}\mathbf{x}(t) such that \begin{aligned} \dot{\mathbf{x}}(t)&=\mathbf{A}(t)\mathbf{x}(t)+\mathbf{b}(t){u}(t)\\ {y}(t)&=\mathbf{c}(t)^{*}\mathbf{x}(t)\end{aligned}~\Longrightarrow~\left\{\begin{aligned} \dot{\mathbf{z}}(t)&=\mathbf{R}\mathbf{z}(t)+\mathbf{P}(t)^{-1}\mathbf{b}(t){u}(t)\\ {y}(t)&=\mathbf{c}(t)^{*}\mathbf{P}(t)\mathbf{z}(t).\end{aligned}\right. (3) A key consequence of this transformation is that the time-dependence in the system has now been isolated in the input/output ports; we term this system structure a port-isolated ltp system . This transformation illuminates conditions for stability for ( 1 ): the original ltp system is asymptotically stable (and hence bibo stable, see e.g., ( Brockett 2015 , §30) ) if and only if \mathbf{R} in ( 3 ) is a stable matrix. Our interest in this transformation is that it allows one to take advantage of powerful model reduction methods that originally were designed for linear time-invariant systems. Although Floquet transformation is not normally viewed as a computational tool, for problems of small to moderate size, algorithms have been developed over the past decade that make effective use of Fourier spectral methods to identify \mathbf{R} and \mathbf{P}(t) in a numerically stable way (See e.g., Deconinck and Kutz 2006 ; Moore 2005 ). These approaches presume access to the monodromy matrix, which makes them intractable for large-scale ltp systems. Our approach builds up truncated Floquet bases (columns of \mathbf{P}(t) ), which will serve for the construction of reduced order models. Floquet transformations are useful yet costly to determine for large-scale ltp systems. The basic transformation follows by noting that the principal fundamental solution matrix, \Phi(t) , for the homogenous ltp system, \dot{\bf x}(t)={\bf A}(t){\bf x}(t) , has special structure: \Phi(t)={\bf P}(t)e^{t\,{\bf R}} where {\bf P}(t)={\bf P}(t+T) is nonsingular and periodic for all t and {\bf R} is a constant matrix. As one may note in ( 3 ), the (time-dependent) basis defined in the columns of {\bf P}(t) can be used to transform a nonhomogeneous ltp system into a corresponding time-invariant system that may be more amenable to analysis (and reduction). However, the computational bottleneck one faces in direct implementation lies in evaluation of the principal fundamental solution matrix over a period, \Phi(t) , followed by (explicit) computation of the matrix logarithm of the monodromy matrix, \Phi(T) . Evaluation of the monodromy

Read the complete text