Learning to Do or Learning While Doing: Reinforcement Learning and Bayesian Optimisation for Online Continuous Tuning

TL;DR

This paper compares reinforcement learning and Bayesian optimisation for online tuning of real-world plants, demonstrating that RL generally performs better but that the choice depends on specific task criteria, aiding autonomous tuning adoption.

Contribution

It provides a comparative analysis of RL and BO in real-world plant tuning, establishing criteria for selecting the appropriate algorithm based on task characteristics.

Findings

RL generally outperforms BO in real-world tuning tasks

The choice of algorithm depends on specific tuning scenario criteria

Guidelines are provided to select the optimal tuning method

Abstract

Online tuning of real-world plants is a complex optimisation problem that continues to require manual intervention by experienced human operators. Autonomous tuning is a rapidly expanding field of research, where learning-based methods, such as Reinforcement Learning-trained Optimisation (RLO) and Bayesian optimisation (BO), hold great promise for achieving outstanding plant performance and reducing tuning times. Which algorithm to choose in different scenarios, however, remains an open question. Here we present a comparative study using a routine task in a real particle accelerator as an example, showing that RLO generally outperforms BO, but is not always the best choice. Based on the study's results, we provide a clear set of criteria to guide the choice of algorithm for a given tuning task. These can ease the adoption of learning-based autonomous tuning solutions to the operation of complex real-world plants, ultimately improving the availability and pushing the limits of operability of these facilities, thereby enabling scientific and engineering advancements.