Communication Dans Un Congrès Année : 2023

Saturated Feedback

Résumé

Plenary 1: Two decades of saturated linear feedback design via global and generalized sector conditions The saturation nonlinearity and its sector properties perhaps provided the main inspiration for absolute stability theory and sector properties. The advent of linear matrix inequalities (LMI) and their efficient solvers, in the mid 1990s, provided fertile ground for the development of LMI-based stability/performance analysis and control design algorithms for linear time-invariant (LTI) plants subject to input saturation. The corresponding results, mainly developed in the past 20 years, started from state feedback designs in parallel with anti-windup augmentations of predefined linear controllers, combined with quadratic Lyapunov certificates ensuring global exponential stability. The peculiar nature of saturation, however, is such that no LTI plant can be globally exponentially stabilized from a bounded input, unless it is already globally exponentially stable in open loop. Due to this fact, non-global (or generalized) sector conditions were invented, which opened up important questions about the size of (suitable estimates of) the basin of attraction. The ensuing generalized (or non-global) absolute stability conditions evolved over the past 15 years into several increasingly advanced control design and performance analysis algorithms for input-saturated systems. In this talk we provide an overview of these developments, covering the notion of generalized sector conditions, highlighting the limits of global versus local designs, the use of deadzone-induced piecewise smooth Lyapunov functions, the multi-agent extensions of the results for bounded synchronization of identical linear plants, and possible approaches for robust designs in the presence of uncertain plants. Plenary 2: Optimizing shifted equilibria for enlarged basins of attraction in quadratically stabilizing saturated feedback With asymmetric actuator saturation standard LMI-based approaches conservatively rely on symmetric sublevel sets of Lyapunov functions and conservative estimates based on the lowest saturation margin, between the available positive and negative actuator range. We describe here an approach based on an asymmetric feedback scheme focusing on shifted equilibria, selected via the solution of an optimization problem ensuring convergence to the origin of the shifted equilibrium. To enable the computational time required by the optimizer, we impose sampled-data updates of the shifted equilibria and cast our description within a hybrid dynamical systems formulation. Exact and inaccurate optimization algorithms are both allowed in our scheme, thus establishing interesting trade-offs between continuous-time dynamics and computationally expensive iterative discrete-time parametric optimization schemes. The results are illustrated on numerical examples and a toy experimental testbed.

Fichier non déposé

Dates et versions

hal-04935683 , version 1 (07-02-2025)

Identifiants

  • HAL Id : hal-04935683 , version 1

Citer

Luca Zaccarian. Saturated Feedback. 42nd Benelux Meeting on Systems and Control, R. Carloni; J. Alonso-Mora; B. Jayawardhana; E. Lefeber, Mar 2023, Elspeet, Netherlands. ⟨hal-04935683⟩
71 Consultations
0 Téléchargements

Partager

  • More