arXiv subjects

Pierre Vuillemin

Publications and source records attributed to Pierre Vuillemin.

13 recordsLinked to original sources

arXiv2025

Tensor-based multivariate function approximation: methods benchmarking and comparison

We evaluate some methods designed for tensor- (or data-) based multivariate model construction (approximation and compression). To this aim, a collection of multivariate functions and an evaluation methodology are suggested. First, these functions, with varying complexity (e.g., number and degree of the variables) and nature (e.g., rational, irrational, differentiable or not, symmetric, etc.) are used to build $n$-dimensional tensors, each of different dimension and memory size. Second, grounded on this tensor, we evaluate the performances of different methods and implementations leading to different types of surrogate models (e.g., rational functions, networks). The accuracy, the computational time, the parameter tuning impact, etc. are monitored and reported. One objective is to evaluate the different available strategies to guide users on the prospects, advantages, and limits of the various tools. The contributions are twofold: (i) to suggest a comprehensive benchmark collection together with a methodology for tensor approximation with a surrogate model and, in addition, (ii) to provide a digest and additional details of the multivariate Loewner Framework (mLF) approach [Antoulas et al., 2025], as well as detailed examples and code.

math.NA↗

arXiv2023

Data-driven port-Hamiltonian structured identification for non-strictly passive systems

In this work, we detail a procedure to construct a reduced order model on the basis of frequency-domain data, that preserves the non-strictly passive property and the port-Hamiltonian structure. The proposed scheme is based on Benner et al. (2020) contribution, which has been adapted (i) to handle non-strictly passive model, and (ii) to handle numerical issues observed when applying the Loewner framework on complex configurations. We validate the proposed scheme on a very complex two-dimensional wave equation, for which the discretized version preserves the port-Hamiltoninan form.

eess.SY↗

arXiv2021

Constructive interpolation points selection in the Loewner framework

This note describes a constructive heuristic to select frequencies of interest within the context of reduced-order modelling by interpolation. The approach is described here through the Loewner framework. Numerical illustrations highlight the benefit it can bring to decrease the required number of interpolation points which is key when the data come from numerically expensive solvers.

math.NA↗

arXiv2021

Performance evaluation of gust load alleviation systems for flexible aircraft via optimal control

The dynamical response of an aircraft subject to gust perturbations is a key element in a preliminary design phase. In particular, the loads induced by gusts along the wing should not exceed some limit values and should even ideally be decreased. Active control is one lever to address this problem. However, evaluating the benefit that active control may bring considering some actuators characteristics or some delay in the loop is a difficult task, especially in the early design phase. This problem is addressed in this paper with an open-loop optimal control framework and more specifically with a direct transcription method resulting in a linear optimisation problem. The approach is illustrated on a realistic aeroelastic aircraft model built with a coupled fluid-structure solver which order is reduced to decrease the number of optimisation variables.

eess.SY↗

arXiv2020

Interpolatory Methods for Generic BizJet Gust Load Alleviation Function

The paper's main contribution concerns the use of interpolatory methods to solve end to end industrial control problems involving complex linear dynamical systems. More in details, contributions show how the rational data and function interpolation framework is a pivotal tool (i) to construct (frequency-limited) reduced order dynamical models appropriate for model-based control design and (ii) to accurately discretise controllers in view of on-board computer-limited implementation. These contributions are illustrated along the paper through the design of an active feedback gust load alleviation function, applied on an industrial generic business jet aircraft use-case. The closed-loop validation and performances evaluation are assessed through the use of an industrial dedicated simulator and considering certification objectives. Although application is centred on aircraft applications, the method is not restrictive and can be extended to any linear dynamical systems.

eess.SY↗

arXiv2020

Mixed interpolatory and inference non-intrusive reduced order modeling with application to pollutants dispersion

On the basis of input-output time-domain data collected from a complex simulator, this paper proposes a constructive methodology to infer a reduced-order linear, bilinear or quadratic time invariant dynamical model reproducing the underlying phenomena. The approach is essentially based on linear dynamical systems and approximation theory. More specifically, it sequentially involves the interpolatory Pencil and Loewner framework, known to be both very versatile and scalable to large-scale data sets, and a linear least square problem involving the raw data and reduced internal variables. With respect to intrusive methods, no prior knowledge on the operator is needed. In addition, compared to the traditional non-intrusive operator inference ones, the proposed approach alleviates the need of measuring the original full-order model internal variables. It is thus applicable to a wider application range than standard intrusive and non-intrusive methods. The rationale is successfully applied on a large eddy simulation of a pollutants dispersion case over an airport area involving multi-scale and multi-physics dynamical phenomena. Despite the simplicity of the resulting low complexity model, the proposed approach shows satisfactory results to predict the pollutants plume pattern while being significantly faster to simulate.

math.DS↗

arXiv2020

Loewner-based Data-driven Iterative Structured Control Design

Stability enforcement remains a challenge in data-driven control paradigms, where no parametrised model of the system is available. For instance, the system's instabilities can be estimated in order to enforce a closed-loop stability constraint on the controller reduction step. In order to avoid this preliminary estimation of instabilities, this paper proposes to embed a closed-loop stability constraint in the design. To that extent, an optimization problem is formulated in order to improve matching between the reference model and the closed-loop while maintaining internal stability. The proposed iterative procedure to solve this problem is illustrated on two numerical examples.

eess.SY↗

arXiv2020

Interpolation-based irrational model control design and stability analysis

The versatility of data-driven approximation by interpolatory methods, originally settled for model approximation purpose, is illustrated in the context of linear controller design and stability analysis of irrational models. To this aim, following an academic driving example described by a linear partial differential equation, it is shown how the Loewner-based interpolation may be an essential ingredient for control design and stability analysis. More specifically, the interpolatory framework is first used to approximate the irrational model by a rational one that can be used for model-based control, and secondly, it is used for direct data-driven control design, showing equivalent results. Finally, this interpolation framework is employed for estimating the stability of the interconnection of the irrational model with a rational controller.

math.OC↗

arXiv2020

Optimal Modal Truncation

This paper revisits the modal truncation from an optimisation point of view. In particular, the concept of dominant poles is formulated with respect to different systems norms as the solution of the associated optimal modal truncation problem. The latter is reformulated as an equivalent convex integer or mixed-integer program. Numerical examples highlight the concept and optimisation approach.

math.OC↗

arXiv2020

Yet another introduction to linear dynamical systems control: From identification and approximation to digital control

This report aims at presenting (yet) a(nother) methodology to design and implement a linear controller for linear dynamical systems on practical applications. The specificity of this report is that authors try to cover (obviously in a non exhaustive way) a wide range of control engineering fields. Indeed, the main purpose is to give a quick overview of standard control engineer approaches to non familiar readers. More specifically, using a simple toy example, we discuss the main steps control engineers usually follow. Namely, (i) the excitation signals construction, (ii) the (continuous-time) linear model construction and approximation, (iii) the (continuous-time) control design, and finally, (iv) its time-domain discretisation and control signal modulation in view of practical implementation. This report is clearly user-oriented and thus focuses on practical aspects (using Matlab code) rather than on theoretical ones, let to the reader's curiosity with few but relevant references.

eess.SY↗

arXiv2019

Technical note: Hybrid Loewner Data Driven Control

This note describes how the Loewner framework can be exploited to create a discrete-time control-law from frequency-data of a continuous-time plant so that their hybrid interconnection matches a given continuous-time reference model up to the Nyquist frequency. The resulting Hybrid Loewner Data Driven Control scheme is illustrated on two numerical examples.

eess.SY↗

arXiv2019

Discretisation of continuous-time linear dynamical model with the Loewner interpolation framework

An interpolation method for discretising continuous-time Linear Time Invariant (LTI) models is proposed in this paper. It consists first in using the Loewner interpolation framework on a specific set of frequency data and secondly to project the resulting model onto a stable subspace. The order of the discretised model may be chosen larger than the initial one thus allowing for trading complexity for accuracy if needed. Numerical examples highlight the efficiency of the method at preserving a satisfactory matching both in magnitude and phase in comparison to standard discretisation methods like ZOH or Tustin.

eess.SY↗

arXiv2012

A Spectral Expression for the Frequency-Limited H2-norm

In this paper, a new simple but yet efficient spectral expression of the frequency-limited H2-norm, denoted H2w-norm, is introduced. The proposed new formulation requires the computation of the system eigenvalues and eigenvectors only, and provides thus an alternative to the well established Gramian-based approach. The interest of this new formulation is in three-folds: (i) it provides a new theoretical framework for the H2w-norm-based optimization approach, such as controller synthesis, filter design and model approximation, (ii) it improves the H2w-norm computation velocity and it applicability to models of higher dimension, and (iii) under some conditions, it allows to handle systems with poles on the imaginary axis. Both mathematical proofs and numerical illustrations are provided to assess this new H2w-norm expression.

eess.SY↗