On the Use of Gradient-Based Solver and Deep Learning Approach in Hierarchical Control: Application to Grand Refrigerators

Abstract

This paper extends the work that has been recently studied on hierarchical control proposed by Alamir et al. (2017). This framework is designed to control interconnecting subsystems such as cryogenic processes or power plants. Based on the previous study, Pham et al. (2022) have shown that handling constraints and non-linearities could dispute the real-time feasibility of the approach. In order to reduce the computation time of the nonlinear model predictive controllers (NMPCs) of the local subsystems, two successful directions are investigated and combined, namely truncated fast gradient based NMPC approach and deep-neural-network-based approach. It is also shown that by doing so, the control updating period can be significantly reduced and the closed-loop performance is greatly improved. This paper can therefore be considered as a concrete implementation and validation of some key ideas in the design of real-time distributed NMPCs. All concepts are validated using the realistic and challenging example of a real cryogenic refrigerator.