Distributed model predictive control of linear systems with coupled constraints based on collective neurodynamic optimization

Publication Type:
Conference Proceeding
Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), 2018, 11320 LNAI pp. 318 - 328
Issue Date:
Full metadata record
© Springer Nature Switzerland AG 2018. Distributed model predictive control explores an array of local predictive controllers that synthesize the control of subsystems independently yet they communicate to efficiently cooperate in achieving the closed-loop control performance. Distributed model predictive control problems naturally result in sequential distributed optimization problems that require real-time solution. This paper presents a collective neurodynamic approach to design and implement the distributed model predictive control of linear systems in the presence of globally coupled constraints. For each subsystem, a neurodynamic model minimizes its cost function using local information only. According to the communication topology of the network, neurodynamic models share information to their neighbours to reach consensus on the optimal control actions to be carried out. The collective neurodynamic models are proven to guarantee the global optimality of the model predictive control system.
Please use this identifier to cite or link to this item: