loading page

Adversarially Robust Bayesian Optimization for Efficient Auto-Tuning of Generic Control Structures under Uncertainty
  • Joel Paulson,
  • Georgios Makrygiorgos,
  • Ali Mesbah
Joel Paulson
The Ohio State University

Corresponding Author:paulson.82@osu.edu

Author Profile
Georgios Makrygiorgos
University of California Berkeley
Author Profile
Ali Mesbah
University of California Berkeley Research
Author Profile

Abstract

The performance of advanced controllers depends on the selection of several tuning parameters that can affect the closed-loop control performance and constraint satisfaction in highly nonlinear and nonconvex ways. There has been a significant interest in auto-tuning of complex control structures using Bayesian optimization (BO). However, an open challenge is how to deal with uncertainties in the closed-loop system that cannot be attributed to a lumped, small-scale noise term. This paper develops an adversarially robust BO (ARBO) method that is suited to auto-tuning problems with significant time-invariant uncertainties in a plant simulator. ARBO uses a Gaussian process model that jointly describes the effect of the tuning parameters and uncertainties on the closed-loop performance. ARBO uses an alternating confidence-bound procedure to simultaneously select the next candidate tuning and uncertainty realizations, implying only one expensive closed-loop simulation is needed at each iteration. The advantages of ARBO are demonstrated on two case studies.
06 Sep 2021Submitted to AIChE Journal
09 Sep 2021Submission Checks Completed
09 Sep 2021Assigned to Editor
13 Sep 2021Reviewer(s) Assigned
30 Nov 2021Editorial Decision: Revise Minor
30 Dec 20211st Revision Received
31 Dec 2021Submission Checks Completed
31 Dec 2021Assigned to Editor
04 Jan 2022Editorial Decision: Accept
Jun 2022Published in AIChE Journal volume 68 issue 6. 10.1002/aic.17591