Robust optimal identification experiment design for multisine excitation - CNRS - Centre national de la recherche scientifique Accéder directement au contenu
Article Dans Une Revue Automatica Année : 2021

Robust optimal identification experiment design for multisine excitation

Résumé

In least costly experiment design, the optimal spectrum of an identification experiment is determined in such a way that the cost of the experiment is minimized under some accuracy constraint on the identified parameter vector. Like all optimal experiment design problems, this optimization problem depends on the unknown true system, which is generally replaced by an initial estimate. One important consequence of this is that we can underestimate the actual cost of the experiment and that the accuracy of the identified model can be lower than desired. Here, based on an a-priori uncertainty set for the true system, we propose a convex optimization approach that allows to prevent these issues from happening. We do this when the to-be-determined spectrum is the one of a multisine signal. 1 Introduction We consider in this paper the problem of optimally designing the spectrum Φ u of the excitation signal u of an open-loop identification experiment. By optimal spectrum , we here mean the spectrum yielding the smallest experiment cost while guaranteeing that the accuracy of the identified parameter vector of the plant transfer function is larger than a given threshold. We thus consider the least costly experiment design framework [5], but the approach can easily be adapted to other (dual) frameworks [10,17,13]. The experiment cost J can be defined as a linear combination of the power of the exci-tation signal u and of the power of the part of the output signal induced by u. The experiment cost will therefore be a function of the spectrum Φ u , but also of the unknown true parameter vector θ 0 (we therefore denote the cost as J (θ 0 , Φ u)). Likewise, the accuracy constraint will also depend on θ 0 and on Φ u since the classical accuracy constraints are of the type P −1 (θ 0 , Φ u) ≥ R adm where P (θ 0 , Φ u) is the covariance matrix of the to-be-identified parameter vector (which depends on θ 0 and Φ u) and R adm a matrix reflecting the desired accuracy. The dependency of the optimal spectrum Φ u,opt on the unknown true parameter vector θ 0 is the so-called chicken-and-egg issue encountered in optimal experiment design. This issue is generally circumvented by replacing θ 0 by
Fichier principal
Vignette du fichier
robustdesignsub.pdf (493.63 Ko) Télécharger le fichier
Origine : Fichiers produits par l'(les) auteur(s)
Loading...

Dates et versions

hal-02363067 , version 1 (14-11-2019)

Identifiants

Citer

Xavier Bombois, Federico Morelli, H. Hjalmarsson, Laurent Bako, Kévin Colin. Robust optimal identification experiment design for multisine excitation. Automatica, 2021, ⟨10.1016/j.automatica.2020.109431⟩. ⟨hal-02363067⟩
146 Consultations
125 Téléchargements

Altmetric

Partager

Gmail Facebook X LinkedIn More