Investigation of fast-NMPC and deep learning approach in fixed-point-based hierarchical control

TL;DR

This paper presents a real-time hierarchical control framework combining fast-NMPC and deep learning to improve computational efficiency and performance in controlling interconnected systems like cryogenic refrigerators.

Contribution

It introduces a novel integration of truncated fast gradient methods and deep neural networks into hierarchical NMPC for real-time applications, validated on a cryogenic refrigerator.

Findings

Reduced control updating period significantly

Enhanced closed-loop performance

Validated approach on a real cryogenic system

Abstract

This paper explores some variations of a hierarchical control framework that has been recently proposed. The framework is dedicated to control a network of interconnected subsystems such as the ones describing cryogenic processes or power plants. Recent investigations showed that handling constraints and nonlinearities might challenge the real-time feasibility of the approach. This paper investigates and combine two successful directions, namely, the use of truncated fast gradient and deep neural networks based controller modeling in order to reduce the computation time of the most critical subsystem. It is also shown that by doing so, the control updating period can be drastically reduced and the closed-loop performances highly improved. The paper can therefore be seen as a concrete implementation and validation of some key ideas in real-time distributed NMPC design. All the concepts are validated using the realistic and challenging example of real-life cryogenic refrigerator.