Towards Gain Tuning for Numerical KKL Observers

IFAC-PapersOnLine, vol. 56, pp. 4061–4067

Abstract

This paper presents a first step towards tuning observers for general nonlinear systems. Relying on recent results around Kazantzis-Kravaris/Luenberger (KKL) observers, we propose an empirical criterion to guide the calibration off the observer, by trading of transient performance and sensitivity to measurement noise. We parametrize the gain matrix and evaluate this criterion over a family of observers for different parameter values. We then use neural networks to learn the mapping between the observer and the nonlinear system, and present a novel method to sample the state-space efficiently for nonlinear regression. We illustrate the merits of this approach in numerical simulations.

Authors 4

1. Mona Buisson-Fenet corresponding Aachen Institute for Data Science in Mechanical Engineering

   Ansys (France) · Centre Automatique et Systèmes · École Nationale Supérieure des Mines de Paris · RWTH Aachen University

   Affiliation as printed

   Ansys Research Team, Ansys France, Villeurbanne, France

   Centre Automatique et Systèmes (CAS), Mines Paris – PSL, Paris, France

   Institute for Data Science in Mechanical Engineering, RWTH Aachen University, Aachen, Germany

2. Lukas Bahr Aachen Institute for Data Science in Mechanical Engineering

   RWTH Aachen University

   Affiliation as printed

   Institute for Data Science in Mechanical Engineering, RWTH Aachen University, Aachen, Germany

3. Valéry Morgenthaler

   Ansys (France)

   Affiliation as printed

   Ansys Research Team, Ansys France, Villeurbanne, France

4. Florent Di Meglio

   Centre Automatique et Systèmes · École Nationale Supérieure des Mines de Paris

   Affiliation as printed

   Centre Automatique et Systèmes (CAS), Mines Paris – PSL, Paris, France