Meta-learning of data-driven controllers with automatic model reference tuning: theory and experimental case study

TL;DR

This paper introduces a meta-learning approach for data-driven control that automates hyperparameter tuning, reducing experimental effort and leveraging prior knowledge, validated through a case study on tuning PI controllers for BLDC motors.

Contribution

It presents a novel meta-learning framework for automatic model reference tuning in data-driven control, improving automation and efficiency over traditional methods.

Findings

Meta-learning reduces tuning time and data requirements.

The approach successfully adapts PI controllers for different BLDC motor configurations.

Experimental results demonstrate improved control performance and ease of tuning.

Abstract

Data-driven control offers a viable option for control scenarios where constructing a system model is expensive or time-consuming. Nonetheless, many of these algorithms are not entirely automated, often necessitating the adjustment of multiple hyperparameters through cumbersome trial-and-error processes and demanding significant amounts of data. In this paper, we explore a meta-learning approach to leverage potentially existing prior knowledge about analogous (though not identical) systems, aiming to reduce both the experimental workload and ease the tuning of the available degrees of freedom. We validate this methodology through an experimental case study involving the tuning of proportional, integral (PI) controllers for brushless DC (BLDC) motors with variable loads and architectures.