MPC Controller Tuning using Bayesian Optimization Techniques

TL;DR

This paper introduces a Bayesian optimization framework to efficiently tune MPC controllers for HVAC systems, significantly reducing the computational effort compared to traditional trial-and-error methods while improving performance.

Contribution

The paper presents a novel BO-based tuning method for MPC in HVAC systems, demonstrating its efficiency and effectiveness through realistic simulations.

Findings

BO finds optimal tuning parameters in 13 simulations

BO reduces computational burden compared to grid search

Tuned parameters lower closed-loop costs

Abstract

We present a Bayesian optimization (BO) framework for tuning model predictive controllers (MPC) of central heating, ventilation, and air conditioning (HVAC) plants. This approach treats the functional relationship between the closed-loop performance of MPC and its tuning parameters as a black-box. The approach is motivated by the observation that evaluating the closed-loop performance of MPC by trial-and-error is time-consuming (e.g., every closed-loop simulation can involve solving thousands of optimization problems). The proposed BO framework seeks to quickly identify the optimal tuning parameters by strategically exploring and exploiting the space of the tuning parameters. The effectiveness of the BO framework is demonstrated by using an MPC controller for a central HVAC plant using realistic data. Here, the BO framework tunes back-off terms for thermal storage tanks to minimize year-long closed-loop costs. Simulation results show that BO can find the optimal back-off terms by conducting 13 year-long simulations, which significantly reduces the computational burden of a naive grid search. We also find that the back-off terms obtained with BO reduce the closed-loop costs.