VABO: Violation-Aware Bayesian Optimization for Closed-Loop Control Performance Optimization with Unmodeled Constraints

TL;DR

This paper introduces VABO, a violation-aware Bayesian optimization algorithm designed to optimize closed-loop control performance while efficiently learning unmodeled constraints, demonstrated on energy minimization in vapor compression systems.

Contribution

VABO is a novel Bayesian optimization method that allows controlled constraint violations to enhance learning and accelerate optimization in systems with unmodeled constraints.

Findings

VABO effectively improves energy efficiency in vapor compression systems.

Allowing budgeted violations accelerates constraint learning and optimization.

VABO outperforms traditional safe and constrained BO methods.

Abstract

We study the problem of performance optimization of closed-loop control systems with unmodeled dynamics. Bayesian optimization (BO) has been demonstrated effective for improving closed-loop performance by automatically tuning controller gains or reference setpoints in a model-free manner. However, BO methods have rarely been tested on dynamical systems with unmodeled constraints. In this paper, we propose a violation-aware BO algorithm (VABO) that optimizes closed-loop performance while simultaneously learning constraint-feasible solutions. Unlike classical constrained BO methods which allow an unlimited constraint violations, or safe BO algorithms that are conservative and try to operate with near-zero violations, we allow budgeted constraint violations to improve constraint learning and accelerate optimization. We demonstrate the effectiveness of our proposed VABO method for energy minimization of industrial vapor compression systems.