Goal-oriented safe active learning for predictive control using Bayesian recurrent neural networks

TL;DR

This paper introduces a goal-oriented safe active learning method for predictive control using Bayesian recurrent neural networks, enabling safe online model adaptation and improved control performance.

Contribution

It proposes a novel online model adaptation scheme within MPC that balances exploration and goal-reaching, with theoretical guarantees and practical validation.

Findings

Achieves economic performance close to full knowledge MPC.

Ensures safety and recursive feasibility during learning.

Progressively improves model accuracy while respecting safety constraints.

Abstract

A key challenge in learning-based model predictive control (MPC) is to collect informative data online for model adaptation while ensuring safety and without penalising control performance. In this paper, we propose an online model adaptation scheme embedded within an MPC framework in which the last-layer parameters of a recurrent neural network are recursively updated via Bayesian learning. This is achieved by means of a goal-oriented safe active learning algorithm that alternates between an exploration phase, where the MPC actively explores system dynamics to collect informative data for model adaptation while still pursuing the main control objective, and a goal-reaching phase, where it focuses exclusively on the main control objective. The algorithm is complemented with theoretical guarantees of (i) recursive feasibility, (ii) safety, (iii) termination of exploration in finite time, and (iv) close-to-optimal performance. Simulation results on a benchmark energy system demonstrate that the proposed framework achieves economic performance comparable to that of an MPC with full system knowledge, while progressively improving model accuracy and respecting operational safety constraints with high probability.