Data-Driven Distributed and Localized Model Predictive Control

Alonso, Carmen Amo; Yang, Fengjun; Matni, Nikolai

doi:10.1109/ojcsys.2022.3171787

Cited by 4 publications

(11 citation statements)

References 36 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Both expressions (11) and ( 9) require the knowledge of all the experimental data, as well as the knowledge of the cost function, which is undesirable for interconnected and possibly large systems. While a distributed data-driven solution to the minimization problem (2) could be obtained using standard techniques for distributed optimization, e.g., see [28], [30], we follow a different approach that will lead to an algorithm with finitetime convergence guarantees. To this aim, we define the following auxiliary problems:…”

Section: Multiagent Learning Of Optimal Controlsmentioning

confidence: 99%

“…Clearly, there is no limit to the number of times that this algorithm can be run, i.e., we can use this approach to design an arbitrarily long controller. A receding horizon implementation is common throughout the literature, often termed Model Predictive Control (MPC), and is also implemented in related works [28], [30]. Comparison of these approaches with our method are discussed next.…”

Section: Receding Horizon Implementationmentioning

confidence: 99%

“…The problem of learning optimal controls in network systems has remained, however, relatively unexplored. To the best of our knowledge, the first result in this direction is [28], soon followed by [29] and [30]. In [28], the authors propose a modified version of the DeePC algorithm [23] to stabilize a network system through a primal-dual flow (while minimizing a quadratic function of the inputs and the states).…”

Section: Introductionmentioning

confidence: 99%

“…This assumption is limiting because it presupposes that the time scales of communication and computation processes are significantly shorter than that of the controller action. An alternative approach appeared in [30], where the same problem is tackled through the system level synthesis framework (SLS) [36]. SLS parameterizes closed-loop system responses directly from collected open-loop trajectories.…”

Section: Introductionmentioning

confidence: 99%

“…SLS parameterizes closed-loop system responses directly from collected open-loop trajectories. The result of [30] reduces to a distributed computation of a dynamic state-feedback controller via the alternating direction method of multipliers (ADMM) [37]. This approach shares much of the limitations of [28], namely the necessity of high frequency communication between nodes in the network, and the need to share input-state trajectories among the networks' nodes.…”

Section: Introductionmentioning

confidence: 99%

See 4 more Smart Citations

Distributed Data-Driven Control of Network Systems

Celi

Baggio

Pasqualetti

2023

IEEE Open J. Control. Syst.

View full text Add to dashboard Cite

Imperfect models lead to imperfect controllers and deriving accurate models from first principles or system identification is especially challenging in networked systems. Instead, data can be used to directly compute controllers, without requiring any system identification or modeling. In this paper we propose a strategy to directly learn control actions when data from past system trajectories is distributed among multiple agents in a network. The approach we develop provably converges to a suboptimal solution in a finite number of steps, bounded by the diameter of the network, and with a sub-optimality gap that can be characterized as a function of data, and that can be made arbitrarily small. We further characterize the robustness properties of our approach and give provable guarantees on its performance when data are affected by noise or by a class of attacks.

show abstract

Section: Multiagent Learning Of Optimal Controlsmentioning

confidence: 99%

Section: Receding Horizon Implementationmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

Distributed Data-Driven Control of Network Systems

Celi

Baggio

Pasqualetti

2023

IEEE Open J. Control. Syst.

View full text Add to dashboard Cite

show abstract

Predictive control of linear discrete-time Markovian jump systems by learning recurrent patterns

Han¹,

Chung²,

Doyle³

2023

Automatica

View full text Add to dashboard Cite

Data-Driven Model Predictive Control With Stability and Robustness Guarantees

Berberich

Köhler

Müller

et al. 2021

IEEE Trans. Automat. Contr.

498

416

View full text Add to dashboard Cite

The development of control methods based on data has seen a surge of interest in recent years. When applying data-driven controllers in real-world applications, providing theoretical guarantees for the closedloop system is of crucial importance to ensure reliable operation. In this review, we provide an overview of data-driven model predictive control (MPC) methods for controlling unknown systems with guarantees on systems-theoretic properties such as stability, robustness, and constraint satisfaction. The considered approaches rely on the Fundamental Lemma from behavioral theory in order to predict input-output trajectories directly from data. We cover various setups, ranging from linear systems and noise-free data to more realistic formulations with noise and nonlinearities, and we provide an overview of different techniques to ensure guarantees for the closed-loop system. Moreover, we discuss avenues for future research that may further improve the theoretical understanding and practical applicability of data-driven MPC.

show abstract

Data-Driven Distributed and Localized Model Predictive Control

Cited by 4 publications

References 36 publications

Distributed Data-Driven Control of Network Systems

Distributed Data-Driven Control of Network Systems

Predictive control of linear discrete-time Markovian jump systems by learning recurrent patterns

Data-Driven Model Predictive Control With Stability and Robustness Guarantees

Contact Info

Product

Resources

About